Skip to content
Snippets Groups Projects

Compare revisions

Changes are shown as if the source revision was being merged into the target revision. Learn more about comparing revisions.

Source

Select target project
No results found

Target

Select target project
  • uba-ki-lab/ubair
1 result
Show changes
Commits on Source (3)
Hide whitespace changes
Inline Side-by-side
Showing
with 2635 additions and 0 deletions
.Rbuildignore 0 → 100644
View file @ 5d21dfb7
^CONTRIBUTING\.md$
^Dockerfile$
^LICENSE\.md$
^Meta$
^README\.Rmd$
^\./inst/extdata/data$
^\.Rproj\.user$
^\.dvc$
^\.gitlab$
^\.gitlab-ci\.yml$
^\.gitlab/issue_templates/.gitkeep$
^\.lintr$
^\.pre-commit-config\.yaml$
^\.venv$
^doc$
^examples$
^renv$
^renv\.lock$
^ubair\.Rproj$
^vignettes/figure/
.Rprofile 0 → 100644
View file @ 5d21dfb7
source("renv/activate.R")
.gitignore 0 → 100644
View file @ 5d21dfb7
.Rproj.user
.Rhistory
.Rdata
.httr-oauth
.DS_Store
.quarto
.venv/
inst/doc
/doc/
/Meta/
CONTRIBUTING.md 0 → 100644
View file @ 5d21dfb7
Contributing to ubair
All contributions to ubair are welcome! This document outlines the process and guidelines for contributing to the project.
Coding Standards
Aim for self-explanatory and readable code to minimize the need for additional documentation.
Reporting Bugs and Feature Requests
Use GitLab issues to report bugs or propose features. While we do not provide a specific template, please include sufficient details to help us understand and address the issue.
Workflow and Version Control
Pull requests (PRs) are welcome. You may either:
• Fork the repository and submit a PR from your fork.
• Work directly on a branch and submit a PR.
Testing and Validation
New features must allow for reproducible results. Validate your code against existing data and workflows to ensure compatibility and consistency.
Licensing
Ensure that any new dependencies added are compatible with the GPL-3.0-or-later used in this project.
General Guidelines
Discuss significant changes (e.g., new features) in a GitLab issue before starting your work. When adding new files or modifying the folder structure, ensure compatibility with the existing structure.
Thank you for contributing to ubair! If you have any questions, feel free to contact the maintainers listed in the README file.
DESCRIPTION 0 → 100644
View file @ 5d21dfb7
Package: ubair
Title: Auswirkungen externer Bedingungen auf die Luftqualität
Version: 1.1.0
Authors@R:
person("Imke", "Voss", , "imke.voss@uba.de", role = c("aut", "cre", "cph"))
person("Raphael", "Franke", , "raphael.franke@uba.de", role = c("aut", "cre"))
Description: Statistische Untersuchung der Auswirkungen externer Bedingungen auf die Luftqualität.
License: GPL (>= 3) + file LICENSE
Depends:
R (>= 4.4.0),
Encoding: UTF-8
Roxygen: list(markdown = TRUE)
RoxygenNote: 7.3.2
Suggests:
testthat (>= 3.0.0),
deepnet,
fastshap,
treeshap,
shapviz,
knitr,
rmarkdown
Config/testthat/edition: 3
Imports:
rlang,
data.table,
dplyr,
ggplot2,
forecast,
lubridate,
tidyr,
yaml,
ranger,
lightgbm
LazyData: true
LazyDataCompression: xz
VignetteBuilder: knitr
Dockerfile 0 → 100644
View file @ 5d21dfb7
FROM rocker/tidyverse:4.4.1
# Set environment variables
ENV R_LIBS_USER=/usr/local/lib/R/site-library
ENV CRAN=https://cran.rstudio.com
# Install system dependencies
RUN apt-get update && apt-get install -y \
python3-pip \
libtool \
automake \
autoconf \
gcc \
g++ \
make \
qpdf \
&& pip install dvc[s3]
# Set the working directory to match GitLab CI
WORKDIR /builds/use-case-luft/ubair
# Copy the project files including renv.lock
COPY . .
# Install R packages via renv and devtools
RUN R -e "install.packages('renv', repos='$CRAN')" && \
R -e "renv::restore()" && \
R -e "install.packages('devtools', repos='$CRAN')"
LICENSE 0 → 100644
View file @ 5d21dfb7
DL-DE->BY-2.0
Data licence Germany – attribution – version 2.0
This licence refers to the sample_data_DESN025 provided in this publication.
Provider of the data: Sächsisches Landesamt für Umwelt, Landwirtschaft und Geologie (LfULG)
Alterations in the data: Codes for incorrect values have been removed.
(1) Any use will be permitted provided it fulfils the requirements of this "Data licence Germany – attribution – Version 2.0".
The data and meta-data provided may, for commercial and non-commercial use, in particular
- be copied, printed, presented, altered, processed and transmitted to third parties;
- be merged with own data and with the data of others and be combined to form new and independent datasets;
- be integrated in internal and external business processes, products and applications in public and non-public electronic networks.
(2) The user must ensure that the source note contains the following information:
- the name of the provider,
- the annotation "Data licence Germany – attribution – Version 2.0" or "dl-de/by-2-0" referring to the licence text available at www.govdata.de/dl-de/by-2-0, and
- a reference to the dataset (URI).
This applies only if the entity keeping the data provides the pieces of information 1-3 for the source note.
(3) Changes, editing, new designs or other amendments must be marked as such in the source note.
URL: http://www.govdata.de/dl-de/by-2-0
LICENSE.md 0 → 100644
View file @ 5d21dfb7
GNU General Public License v3.0 or later
========================================
_Version 3, 29 June 2007_
_Copyright © 2007 Free Software Foundation, Inc. &lt;<http://fsf.org/>&gt;_
Everyone is permitted to copy and distribute verbatim copies of this license
document, but changing it is not allowed.
## Preamble
The GNU General Public License is a free, copyleft license for software and other
kinds of works.
The licenses for most software and other practical works are designed to take away
your freedom to share and change the works. By contrast, the GNU General Public
License is intended to guarantee your freedom to share and change all versions of a
program--to make sure it remains free software for all its users. We, the Free
Software Foundation, use the GNU General Public License for most of our software; it
applies also to any other work released this way by its authors. You can apply it to
your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General
Public Licenses are designed to make sure that you have the freedom to distribute
copies of free software (and charge for them if you wish), that you receive source
code or can get it if you want it, that you can change the software or use pieces of
it in new free programs, and that you know you can do these things.
To protect your rights, we need to prevent others from denying you these rights or
asking you to surrender the rights. Therefore, you have certain responsibilities if
you distribute copies of the software, or if you modify it: responsibilities to
respect the freedom of others.
For example, if you distribute copies of such a program, whether gratis or for a fee,
you must pass on to the recipients the same freedoms that you received. You must make
sure that they, too, receive or can get the source code. And you must show them these
terms so they know their rights.
Developers that use the GNU GPL protect your rights with two steps: **(1)** assert
copyright on the software, and **(2)** offer you this License giving you legal permission
to copy, distribute and/or modify it.
For the developers' and authors' protection, the GPL clearly explains that there is
no warranty for this free software. For both users' and authors' sake, the GPL
requires that modified versions be marked as changed, so that their problems will not
be attributed erroneously to authors of previous versions.
Some devices are designed to deny users access to install or run modified versions of
the software inside them, although the manufacturer can do so. This is fundamentally
incompatible with the aim of protecting users' freedom to change the software. The
systematic pattern of such abuse occurs in the area of products for individuals to
use, which is precisely where it is most unacceptable. Therefore, we have designed
this version of the GPL to prohibit the practice for those products. If such problems
arise substantially in other domains, we stand ready to extend this provision to
those domains in future versions of the GPL, as needed to protect the freedom of
users.
Finally, every program is threatened constantly by software patents. States should
not allow patents to restrict development and use of software on general-purpose
computers, but in those that do, we wish to avoid the special danger that patents
applied to a free program could make it effectively proprietary. To prevent this, the
GPL assures that patents cannot be used to render the program non-free.
The precise terms and conditions for copying, distribution and modification follow.
## TERMS AND CONDITIONS
### 0. Definitions
“This License” refers to version 3 of the GNU General Public License.
“Copyright” also means copyright-like laws that apply to other kinds of
works, such as semiconductor masks.
“The Program” refers to any copyrightable work licensed under this
License. Each licensee is addressed as “you”. “Licensees” and
“recipients” may be individuals or organizations.
To “modify” a work means to copy from or adapt all or part of the work in
a fashion requiring copyright permission, other than the making of an exact copy. The
resulting work is called a “modified version” of the earlier work or a
work “based on” the earlier work.
A “covered work” means either the unmodified Program or a work based on
the Program.
To “propagate” a work means to do anything with it that, without
permission, would make you directly or secondarily liable for infringement under
applicable copyright law, except executing it on a computer or modifying a private
copy. Propagation includes copying, distribution (with or without modification),
making available to the public, and in some countries other activities as well.
To “convey” a work means any kind of propagation that enables other
parties to make or receive copies. Mere interaction with a user through a computer
network, with no transfer of a copy, is not conveying.
An interactive user interface displays “Appropriate Legal Notices” to the
extent that it includes a convenient and prominently visible feature that **(1)**
displays an appropriate copyright notice, and **(2)** tells the user that there is no
warranty for the work (except to the extent that warranties are provided), that
licensees may convey the work under this License, and how to view a copy of this
License. If the interface presents a list of user commands or options, such as a
menu, a prominent item in the list meets this criterion.
### 1. Source Code
The “source code” for a work means the preferred form of the work for
making modifications to it. “Object code” means any non-source form of a
work.
A “Standard Interface” means an interface that either is an official
standard defined by a recognized standards body, or, in the case of interfaces
specified for a particular programming language, one that is widely used among
developers working in that language.
The “System Libraries” of an executable work include anything, other than
the work as a whole, that **(a)** is included in the normal form of packaging a Major
Component, but which is not part of that Major Component, and **(b)** serves only to
enable use of the work with that Major Component, or to implement a Standard
Interface for which an implementation is available to the public in source code form.
A “Major Component”, in this context, means a major essential component
(kernel, window system, and so on) of the specific operating system (if any) on which
the executable work runs, or a compiler used to produce the work, or an object code
interpreter used to run it.
The “Corresponding Source” for a work in object code form means all the
source code needed to generate, install, and (for an executable work) run the object
code and to modify the work, including scripts to control those activities. However,
it does not include the work's System Libraries, or general-purpose tools or
generally available free programs which are used unmodified in performing those
activities but which are not part of the work. For example, Corresponding Source
includes interface definition files associated with source files for the work, and
the source code for shared libraries and dynamically linked subprograms that the work
is specifically designed to require, such as by intimate data communication or
control flow between those subprograms and other parts of the work.
The Corresponding Source need not include anything that users can regenerate
automatically from other parts of the Corresponding Source.
The Corresponding Source for a work in source code form is that same work.
### 2. Basic Permissions
All rights granted under this License are granted for the term of copyright on the
Program, and are irrevocable provided the stated conditions are met. This License
explicitly affirms your unlimited permission to run the unmodified Program. The
output from running a covered work is covered by this License only if the output,
given its content, constitutes a covered work. This License acknowledges your rights
of fair use or other equivalent, as provided by copyright law.
You may make, run and propagate covered works that you do not convey, without
conditions so long as your license otherwise remains in force. You may convey covered
works to others for the sole purpose of having them make modifications exclusively
for you, or provide you with facilities for running those works, provided that you
comply with the terms of this License in conveying all material for which you do not
control copyright. Those thus making or running the covered works for you must do so
exclusively on your behalf, under your direction and control, on terms that prohibit
them from making any copies of your copyrighted material outside their relationship
with you.
Conveying under any other circumstances is permitted solely under the conditions
stated below. Sublicensing is not allowed; section 10 makes it unnecessary.
### 3. Protecting Users' Legal Rights From Anti-Circumvention Law
No covered work shall be deemed part of an effective technological measure under any
applicable law fulfilling obligations under article 11 of the WIPO copyright treaty
adopted on 20 December 1996, or similar laws prohibiting or restricting circumvention
of such measures.
When you convey a covered work, you waive any legal power to forbid circumvention of
technological measures to the extent such circumvention is effected by exercising
rights under this License with respect to the covered work, and you disclaim any
intention to limit operation or modification of the work as a means of enforcing,
against the work's users, your or third parties' legal rights to forbid circumvention
of technological measures.
### 4. Conveying Verbatim Copies
You may convey verbatim copies of the Program's source code as you receive it, in any
medium, provided that you conspicuously and appropriately publish on each copy an
appropriate copyright notice; keep intact all notices stating that this License and
any non-permissive terms added in accord with section 7 apply to the code; keep
intact all notices of the absence of any warranty; and give all recipients a copy of
this License along with the Program.
You may charge any price or no price for each copy that you convey, and you may offer
support or warranty protection for a fee.
### 5. Conveying Modified Source Versions
You may convey a work based on the Program, or the modifications to produce it from
the Program, in the form of source code under the terms of section 4, provided that
you also meet all of these conditions:
* **a)** The work must carry prominent notices stating that you modified it, and giving a
relevant date.
* **b)** The work must carry prominent notices stating that it is released under this
License and any conditions added under section 7. This requirement modifies the
requirement in section 4 to “keep intact all notices”.
* **c)** You must license the entire work, as a whole, under this License to anyone who
comes into possession of a copy. This License will therefore apply, along with any
applicable section 7 additional terms, to the whole of the work, and all its parts,
regardless of how they are packaged. This License gives no permission to license the
work in any other way, but it does not invalidate such permission if you have
separately received it.
* **d)** If the work has interactive user interfaces, each must display Appropriate Legal
Notices; however, if the Program has interactive interfaces that do not display
Appropriate Legal Notices, your work need not make them do so.
A compilation of a covered work with other separate and independent works, which are
not by their nature extensions of the covered work, and which are not combined with
it such as to form a larger program, in or on a volume of a storage or distribution
medium, is called an “aggregate” if the compilation and its resulting
copyright are not used to limit the access or legal rights of the compilation's users
beyond what the individual works permit. Inclusion of a covered work in an aggregate
does not cause this License to apply to the other parts of the aggregate.
### 6. Conveying Non-Source Forms
You may convey a covered work in object code form under the terms of sections 4 and
5, provided that you also convey the machine-readable Corresponding Source under the
terms of this License, in one of these ways:
* **a)** Convey the object code in, or embodied in, a physical product (including a
physical distribution medium), accompanied by the Corresponding Source fixed on a
durable physical medium customarily used for software interchange.
* **b)** Convey the object code in, or embodied in, a physical product (including a
physical distribution medium), accompanied by a written offer, valid for at least
three years and valid for as long as you offer spare parts or customer support for
that product model, to give anyone who possesses the object code either **(1)** a copy of
the Corresponding Source for all the software in the product that is covered by this
License, on a durable physical medium customarily used for software interchange, for
a price no more than your reasonable cost of physically performing this conveying of
source, or **(2)** access to copy the Corresponding Source from a network server at no
charge.
* **c)** Convey individual copies of the object code with a copy of the written offer to
provide the Corresponding Source. This alternative is allowed only occasionally and
noncommercially, and only if you received the object code with such an offer, in
accord with subsection 6b.
* **d)** Convey the object code by offering access from a designated place (gratis or for
a charge), and offer equivalent access to the Corresponding Source in the same way
through the same place at no further charge. You need not require recipients to copy
the Corresponding Source along with the object code. If the place to copy the object
code is a network server, the Corresponding Source may be on a different server
(operated by you or a third party) that supports equivalent copying facilities,
provided you maintain clear directions next to the object code saying where to find
the Corresponding Source. Regardless of what server hosts the Corresponding Source,
you remain obligated to ensure that it is available for as long as needed to satisfy
these requirements.
* **e)** Convey the object code using peer-to-peer transmission, provided you inform
other peers where the object code and Corresponding Source of the work are being
offered to the general public at no charge under subsection 6d.
A separable portion of the object code, whose source code is excluded from the
Corresponding Source as a System Library, need not be included in conveying the
object code work.
A “User Product” is either **(1)** a “consumer product”, which
means any tangible personal property which is normally used for personal, family, or
household purposes, or **(2)** anything designed or sold for incorporation into a
dwelling. In determining whether a product is a consumer product, doubtful cases
shall be resolved in favor of coverage. For a particular product received by a
particular user, “normally used” refers to a typical or common use of
that class of product, regardless of the status of the particular user or of the way
in which the particular user actually uses, or expects or is expected to use, the
product. A product is a consumer product regardless of whether the product has
substantial commercial, industrial or non-consumer uses, unless such uses represent
the only significant mode of use of the product.
“Installation Information” for a User Product means any methods,
procedures, authorization keys, or other information required to install and execute
modified versions of a covered work in that User Product from a modified version of
its Corresponding Source. The information must suffice to ensure that the continued
functioning of the modified object code is in no case prevented or interfered with
solely because modification has been made.
If you convey an object code work under this section in, or with, or specifically for
use in, a User Product, and the conveying occurs as part of a transaction in which
the right of possession and use of the User Product is transferred to the recipient
in perpetuity or for a fixed term (regardless of how the transaction is
characterized), the Corresponding Source conveyed under this section must be
accompanied by the Installation Information. But this requirement does not apply if
neither you nor any third party retains the ability to install modified object code
on the User Product (for example, the work has been installed in ROM).
The requirement to provide Installation Information does not include a requirement to
continue to provide support service, warranty, or updates for a work that has been
modified or installed by the recipient, or for the User Product in which it has been
modified or installed. Access to a network may be denied when the modification itself
materially and adversely affects the operation of the network or violates the rules
and protocols for communication across the network.
Corresponding Source conveyed, and Installation Information provided, in accord with
this section must be in a format that is publicly documented (and with an
implementation available to the public in source code form), and must require no
special password or key for unpacking, reading or copying.
### 7. Additional Terms
“Additional permissions” are terms that supplement the terms of this
License by making exceptions from one or more of its conditions. Additional
permissions that are applicable to the entire Program shall be treated as though they
were included in this License, to the extent that they are valid under applicable
law. If additional permissions apply only to part of the Program, that part may be
used separately under those permissions, but the entire Program remains governed by
this License without regard to the additional permissions.
When you convey a copy of a covered work, you may at your option remove any
additional permissions from that copy, or from any part of it. (Additional
permissions may be written to require their own removal in certain cases when you
modify the work.) You may place additional permissions on material, added by you to a
covered work, for which you have or can give appropriate copyright permission.
Notwithstanding any other provision of this License, for material you add to a
covered work, you may (if authorized by the copyright holders of that material)
supplement the terms of this License with terms:
* **a)** Disclaiming warranty or limiting liability differently from the terms of
sections 15 and 16 of this License; or
* **b)** Requiring preservation of specified reasonable legal notices or author
attributions in that material or in the Appropriate Legal Notices displayed by works
containing it; or
* **c)** Prohibiting misrepresentation of the origin of that material, or requiring that
modified versions of such material be marked in reasonable ways as different from the
original version; or
* **d)** Limiting the use for publicity purposes of names of licensors or authors of the
material; or
* **e)** Declining to grant rights under trademark law for use of some trade names,
trademarks, or service marks; or
* **f)** Requiring indemnification of licensors and authors of that material by anyone
who conveys the material (or modified versions of it) with contractual assumptions of
liability to the recipient, for any liability that these contractual assumptions
directly impose on those licensors and authors.
All other non-permissive additional terms are considered “further
restrictions” within the meaning of section 10. If the Program as you received
it, or any part of it, contains a notice stating that it is governed by this License
along with a term that is a further restriction, you may remove that term. If a
license document contains a further restriction but permits relicensing or conveying
under this License, you may add to a covered work material governed by the terms of
that license document, provided that the further restriction does not survive such
relicensing or conveying.
If you add terms to a covered work in accord with this section, you must place, in
the relevant source files, a statement of the additional terms that apply to those
files, or a notice indicating where to find the applicable terms.
Additional terms, permissive or non-permissive, may be stated in the form of a
separately written license, or stated as exceptions; the above requirements apply
either way.
### 8. Termination
You may not propagate or modify a covered work except as expressly provided under
this License. Any attempt otherwise to propagate or modify it is void, and will
automatically terminate your rights under this License (including any patent licenses
granted under the third paragraph of section 11).
However, if you cease all violation of this License, then your license from a
particular copyright holder is reinstated **(a)** provisionally, unless and until the
copyright holder explicitly and finally terminates your license, and **(b)** permanently,
if the copyright holder fails to notify you of the violation by some reasonable means
prior to 60 days after the cessation.
Moreover, your license from a particular copyright holder is reinstated permanently
if the copyright holder notifies you of the violation by some reasonable means, this
is the first time you have received notice of violation of this License (for any
work) from that copyright holder, and you cure the violation prior to 30 days after
your receipt of the notice.
Termination of your rights under this section does not terminate the licenses of
parties who have received copies or rights from you under this License. If your
rights have been terminated and not permanently reinstated, you do not qualify to
receive new licenses for the same material under section 10.
### 9. Acceptance Not Required for Having Copies
You are not required to accept this License in order to receive or run a copy of the
Program. Ancillary propagation of a covered work occurring solely as a consequence of
using peer-to-peer transmission to receive a copy likewise does not require
acceptance. However, nothing other than this License grants you permission to
propagate or modify any covered work. These actions infringe copyright if you do not
accept this License. Therefore, by modifying or propagating a covered work, you
indicate your acceptance of this License to do so.
### 10. Automatic Licensing of Downstream Recipients
Each time you convey a covered work, the recipient automatically receives a license
from the original licensors, to run, modify and propagate that work, subject to this
License. You are not responsible for enforcing compliance by third parties with this
License.
An “entity transaction” is a transaction transferring control of an
organization, or substantially all assets of one, or subdividing an organization, or
merging organizations. If propagation of a covered work results from an entity
transaction, each party to that transaction who receives a copy of the work also
receives whatever licenses to the work the party's predecessor in interest had or
could give under the previous paragraph, plus a right to possession of the
Corresponding Source of the work from the predecessor in interest, if the predecessor
has it or can get it with reasonable efforts.
You may not impose any further restrictions on the exercise of the rights granted or
affirmed under this License. For example, you may not impose a license fee, royalty,
or other charge for exercise of rights granted under this License, and you may not
initiate litigation (including a cross-claim or counterclaim in a lawsuit) alleging
that any patent claim is infringed by making, using, selling, offering for sale, or
importing the Program or any portion of it.
### 11. Patents
A “contributor” is a copyright holder who authorizes use under this
License of the Program or a work on which the Program is based. The work thus
licensed is called the contributor's “contributor version”.
A contributor's “essential patent claims” are all patent claims owned or
controlled by the contributor, whether already acquired or hereafter acquired, that
would be infringed by some manner, permitted by this License, of making, using, or
selling its contributor version, but do not include claims that would be infringed
only as a consequence of further modification of the contributor version. For
purposes of this definition, “control” includes the right to grant patent
sublicenses in a manner consistent with the requirements of this License.
Each contributor grants you a non-exclusive, worldwide, royalty-free patent license
under the contributor's essential patent claims, to make, use, sell, offer for sale,
import and otherwise run, modify and propagate the contents of its contributor
version.
In the following three paragraphs, a “patent license” is any express
agreement or commitment, however denominated, not to enforce a patent (such as an
express permission to practice a patent or covenant not to sue for patent
infringement). To “grant” such a patent license to a party means to make
such an agreement or commitment not to enforce a patent against the party.
If you convey a covered work, knowingly relying on a patent license, and the
Corresponding Source of the work is not available for anyone to copy, free of charge
and under the terms of this License, through a publicly available network server or
other readily accessible means, then you must either **(1)** cause the Corresponding
Source to be so available, or **(2)** arrange to deprive yourself of the benefit of the
patent license for this particular work, or **(3)** arrange, in a manner consistent with
the requirements of this License, to extend the patent license to downstream
recipients. “Knowingly relying” means you have actual knowledge that, but
for the patent license, your conveying the covered work in a country, or your
recipient's use of the covered work in a country, would infringe one or more
identifiable patents in that country that you have reason to believe are valid.
If, pursuant to or in connection with a single transaction or arrangement, you
convey, or propagate by procuring conveyance of, a covered work, and grant a patent
license to some of the parties receiving the covered work authorizing them to use,
propagate, modify or convey a specific copy of the covered work, then the patent
license you grant is automatically extended to all recipients of the covered work and
works based on it.
A patent license is “discriminatory” if it does not include within the
scope of its coverage, prohibits the exercise of, or is conditioned on the
non-exercise of one or more of the rights that are specifically granted under this
License. You may not convey a covered work if you are a party to an arrangement with
a third party that is in the business of distributing software, under which you make
payment to the third party based on the extent of your activity of conveying the
work, and under which the third party grants, to any of the parties who would receive
the covered work from you, a discriminatory patent license **(a)** in connection with
copies of the covered work conveyed by you (or copies made from those copies), or **(b)**
primarily for and in connection with specific products or compilations that contain
the covered work, unless you entered into that arrangement, or that patent license
was granted, prior to 28 March 2007.
Nothing in this License shall be construed as excluding or limiting any implied
license or other defenses to infringement that may otherwise be available to you
under applicable patent law.
### 12. No Surrender of Others' Freedom
If conditions are imposed on you (whether by court order, agreement or otherwise)
that contradict the conditions of this License, they do not excuse you from the
conditions of this License. If you cannot convey a covered work so as to satisfy
simultaneously your obligations under this License and any other pertinent
obligations, then as a consequence you may not convey it at all. For example, if you
agree to terms that obligate you to collect a royalty for further conveying from
those to whom you convey the Program, the only way you could satisfy both those terms
and this License would be to refrain entirely from conveying the Program.
### 13. Use with the GNU Affero General Public License
Notwithstanding any other provision of this License, you have permission to link or
combine any covered work with a work licensed under version 3 of the GNU Affero
General Public License into a single combined work, and to convey the resulting work.
The terms of this License will continue to apply to the part which is the covered
work, but the special requirements of the GNU Affero General Public License, section
13, concerning interaction through a network will apply to the combination as such.
### 14. Revised Versions of this License
The Free Software Foundation may publish revised and/or new versions of the GNU
General Public License from time to time. Such new versions will be similar in spirit
to the present version, but may differ in detail to address new problems or concerns.
Each version is given a distinguishing version number. If the Program specifies that
a certain numbered version of the GNU General Public License “or any later
version” applies to it, you have the option of following the terms and
conditions either of that numbered version or of any later version published by the
Free Software Foundation. If the Program does not specify a version number of the GNU
General Public License, you may choose any version ever published by the Free
Software Foundation.
If the Program specifies that a proxy can decide which future versions of the GNU
General Public License can be used, that proxy's public statement of acceptance of a
version permanently authorizes you to choose that version for the Program.
Later license versions may give you additional or different permissions. However, no
additional obligations are imposed on any author or copyright holder as a result of
your choosing to follow a later version.
### 15. Disclaimer of Warranty
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.
EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
PROVIDE THE PROGRAM “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER
EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE
QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE
DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
### 16. Limitation of Liability
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY
COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS THE PROGRAM AS
PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL,
INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE
PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE
OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE
WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
POSSIBILITY OF SUCH DAMAGES.
### 17. Interpretation of Sections 15 and 16
If the disclaimer of warranty and limitation of liability provided above cannot be
given local legal effect according to their terms, reviewing courts shall apply local
law that most closely approximates an absolute waiver of all civil liability in
connection with the Program, unless a warranty or assumption of liability accompanies
a copy of the Program in return for a fee.
_END OF TERMS AND CONDITIONS_
## How to Apply These Terms to Your New Programs
If you develop a new program, and you want it to be of the greatest possible use to
the public, the best way to achieve this is to make it free software which everyone
can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest to attach them
to the start of each source file to most effectively state the exclusion of warranty;
and each file should have at least the “copyright” line and a pointer to
where the full notice is found.
<one line to give the program's name and a brief idea of what it does.>
Copyright (C) <year> <name of author>
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
Also add information on how to contact you by electronic and paper mail.
If the program does terminal interaction, make it output a short notice like this
when it starts in an interactive mode:
<program> Copyright (C) <year> <name of author>
This program comes with ABSOLUTELY NO WARRANTY; for details type 'show w'.
This is free software, and you are welcome to redistribute it
under certain conditions; type 'show c' for details.
The hypothetical commands `show w` and `show c` should show the appropriate parts of
the General Public License. Of course, your program's commands might be different;
for a GUI interface, you would use an “about box”.
You should also get your employer (if you work as a programmer) or school, if any, to
sign a “copyright disclaimer” for the program, if necessary. For more
information on this, and how to apply and follow the GNU GPL, see
&lt;<http://www.gnu.org/licenses/>&gt;.
The GNU General Public License does not permit incorporating your program into
proprietary programs. If your program is a subroutine library, you may consider it
more useful to permit linking proprietary applications with the library. If this is
what you want to do, use the GNU Lesser General Public License instead of this
License. But first, please read
&lt;<http://www.gnu.org/philosophy/why-not-lgpl.html>&gt;.
NAMESPACE 0 → 100644
View file @ 5d21dfb7
# Generated by roxygen2: do not edit by hand
export(calc_performance_metrics)
export(calc_summary_statistics)
export(clean_data)
export(copy_default_params)
export(detrend)
export(estimate_effect_size)
export(get_meteo_available)
export(load_params)
export(load_uba_data_from_dir)
export(plot_counterfactual)
export(plot_station_measurements)
export(prepare_data_for_modelling)
export(rescale_predictions)
export(retrend_predictions)
export(run_counterfactual)
export(run_dynamic_regression)
export(run_fnn)
export(run_lightgbm)
export(run_rf)
export(scale_data)
export(split_data_counterfactual)
import(lightgbm)
import(ranger)
import(splines)
importFrom(data.table,":=")
importFrom(data.table,.SD)
importFrom(data.table,melt)
importFrom(dplyr,"%>%")
importFrom(dplyr,across)
importFrom(dplyr,mutate)
importFrom(dplyr,select)
importFrom(dplyr,where)
importFrom(forecast,BoxCox.lambda)
importFrom(forecast,auto.arima)
importFrom(forecast,forecast)
importFrom(ggplot2,aes)
importFrom(ggplot2,facet_wrap)
importFrom(ggplot2,geom_line)
importFrom(ggplot2,geom_ribbon)
importFrom(ggplot2,geom_smooth)
importFrom(ggplot2,geom_vline)
importFrom(ggplot2,ggplot)
importFrom(ggplot2,labs)
importFrom(ggplot2,theme_bw)
importFrom(lubridate,ymd_h)
importFrom(lubridate,ymd_hm)
importFrom(stats,lm)
importFrom(stats,predict)
importFrom(stats,spline)
importFrom(tidyr,gather)
importFrom(yaml,read_yaml)
NEWS.md 0 → 100644
View file @ 5d21dfb7
## ubair 1.1.0
**Initial CRAN Submission**\
**Public Release**
- First public release of `ubair` on CRAN.
- Previous internal versions (pre-CRAN) were used privately or shared within a specific group.
- The package is intended for use with air quality station measurement data. These can be obtained for Germany via [Air Quality Data API (UBA)](https://www.umweltbundesamt.de/daten/luft/luftdaten/doc) or per request to immission@uba.de
- Includes the following features:
- **Counterfactual analysis**: Apply counterfactual analysis to assess external effects on air quality based on meteorological data, with support for four models: LightGBM, Random Forest, Feedforward Neural Network, and Dynamic Regression.
- **Data detrending and retrending**: Enable users to remove and reintroduce trends in data as needed.
- **Visualization and evaluation**: Provide tools for visualizing and evaluating the results of counterfactual analyses.
- **Data preprocessing**: Preprocess data to prepare it for counterfactual analysis.
R/counterfactual_model.R 0 → 100644
View file @ 5d21dfb7
#' Full counterfactual simulation run
#'
#' Chains detrending, training of a selected model, prediction and retrending together
#' for ease of use. See documentation of individual functions for details.
#' @param split_data List of two named dataframes called train and apply
#' @param params A list of parameters that define the following:
#' \describe{
#' \item{meteo_variables}{A character vector specifying the names of the
#' meteorological variables used as inputs.}
#' \item{model}{A list of hyperparameters for training the chosen model. Name of this list
#' and its parameters depend on the chosen models. See [ubair::run_dynamic_regression()],
#' [ubair::run_lightgbm()], [ubair::run_rf()] and [ubair::run_fnn()] functions for details}
#' }
#' @param detrending_function String which defines type of trend to remove.
#' Options are "linear","quadratic", "exponential", "spline", "none". See [ubair::detrend()]
#' and [ubair::retrend_predictions()] for details.
#' @param model_type String to decide which model to use. Current options random
#' forest "rf", gradient boosted decision trees "lightgbm", "dynamic_regression" and feedforward neural network "fnn"
#' @param alpha Confidence level of the prediction interval between 0 and 1.
#' @param log_transform If TRUE, uses log transformation during detrending and
#' retrending. For details see [ubair::detrend()] documentation
#' @param calc_shaps Boolean value. If TRUE, calculate SHAP values for the
#' method used and format them so they can be visualised with \code{\link[shapviz:sv_importance]{shapviz:sv_importance()}} and
#' \code{\link[shapviz:sv_dependence]{shapviz:sv_dependence()}}.
#' The SHAP values are generated for a subset (or all, depending on the size of the dataset) of the
#' test data.
#' @return Data frame of predictions and model
#' @examples
#' \dontrun{
#' split_data <- split_data_counterfactual(
#' dt_prepared, training_start,
#' training_end, application_start, application_end
#' )
#' res <- run_counterfactual(split_data, params, detrending_function = "linear")
#' prediction <- res$retrended_predictions
#' random_forest_model <- res$model
#' }
#' @export
run_counterfactual <- function(split_data,
params,
detrending_function = "none",
model_type = "rf",
alpha = 0.9,
log_transform = FALSE,
calc_shaps = FALSE) {
variables <- c("day_julian", "weekday", "hour", params$meteo_variables)
detrended_list <- detrend(
split_data,
mode = detrending_function,
log_transform = log_transform
)
trend <- detrended_list$model
detrended_train <- detrended_list$train
detrended_apply <- detrended_list$apply
if (model_type == "rf") {
detrended_train <- detrended_train %>% select(value, !!variables)
res <- run_rf(
train = detrended_train,
test = detrended_apply,
model_params = params$random_forest,
alpha = alpha,
calc_shaps = calc_shaps
)
} else if (model_type == "lightgbm") {
detrended_train <- detrended_train %>%
select(value, dplyr::any_of(variables))
res <- run_lightgbm(
train = detrended_train,
test = detrended_apply,
model_params = params$lightgbm,
alpha = alpha,
calc_shaps = calc_shaps
)
} else if (model_type == "dynamic_regression") {
res <- run_dynamic_regression(
train = detrended_train,
test = detrended_apply,
params = params,
alpha = alpha,
calc_shaps = calc_shaps
)
} else if (model_type == "fnn") {
res <- suppressMessages(run_fnn(
train = detrended_train,
test = detrended_apply,
params = params,
calc_shaps = calc_shaps
))
} else {
stop("Wrong model_type. Select one of 'rf', 'lightgbm', 'dynamic_regression', 'fnn'.")
}
retrended_predictions <- retrend_predictions(res$dt_predictions,
trend,
log_transform = log_transform
)
list(prediction = retrended_predictions, model = res$model, importance = res$importance)
}
#' Run the dynamic regression model
#'
#' This function trains a dynamic regression model with fourier transformed temporal features
#' and meteorological variables as external regressors on the
#' specified training dataset and makes predictions on the test dataset in a
#' counterfactual scenario. This is referred to as a dynamic regression model in
#' [Forecasting: Principles and Practise, Chapter 10 - Dynamic regression models](https://otexts.com/fpp3/dynamic.html)
#'
#' Note: Runs the dynamic regression model for individualised use with own data pipeline.
#' Otherwise use [ubair::run_counterfactual()] to call this function.
#' @param train Dataframe of train data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param test Dataframe of test data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param params list of hyperparameters to use in dynamic_regression call. Only uses ntrain to specify
#' the number of data points to use for training. Default is 8760 which results in
#' 1 year of hourly data
#' @param alpha Confidence level of the prediction interval between 0 and 1.
#' @param calc_shaps Boolean value. If TRUE, calculate SHAP values for the
#' method used and format them so they can be visualised with \code{\link[shapviz:sv_importance]{shapviz:sv_importance()}} and
#' \code{\link[shapviz:sv_dependence]{shapviz:sv_dependence()}}.
#' The SHAP values are generated for a subset (or all, depending on the size of the dataset) of the
#' test data.
#' @return Data frame of predictions and model
#' @importFrom forecast BoxCox.lambda auto.arima forecast
#' @export
run_dynamic_regression <- function(train,
test,
params,
alpha,
calc_shaps) {
train <- train %>% dplyr::filter(date < test$date[1])
# 24 * 365 = 8760 (1 year of training data)
ntrain <- ifelse(is.null(params$dynamic_regression$ntrain), 8760, params$dynamic_regression$ntrain)
train <- utils::tail(train, ntrain)
message(
paste("Using data for dynamic regression training from ", min(train$date), "to ", max(train$date)),
". Too long training series can lead to worse performance. Adjust this via the dynamic_regression$ntrain hyperparameter."
)
train_transformed <- .transform_input(train, params)
test_transformed <- .transform_input(test, params)
scale_result <- scale_data(
train_data = train_transformed,
apply_data = test_transformed
)
xreg <- scale_result$train %>%
dplyr::select(-value) %>%
as.matrix()
xreg_pred <- scale_result$apply %>%
dplyr::select(-value) %>%
as.matrix()
y <- scale_result$train$value %>% stats::ts()
model <- forecast::auto.arima(y,
d = 0, xreg = xreg,
seasonal = FALSE,
trace = FALSE,
allowdrift = TRUE,
allowmean = TRUE,
lambda = NULL,
biasadj = TRUE
)
pred <- forecast::forecast(
object = model,
xreg = xreg_pred,
level = alpha,
lambda = NULL,
biasadj = TRUE
)
test$prediction <- pred$mean %>% as.numeric()
test$prediction_lower <- pred$lower[, 1] %>% as.numeric()
test$prediction_upper <- pred$upper[, 1] %>% as.numeric()
dt_predictions <- rescale_predictions(
scale_result = scale_result,
dt_predictions = test
)
if (calc_shaps) {
rlang::check_installed(c("fastshap", "shapviz"),
reason = "calculate shap values for dynamic regression"
)
shap_vals <- fastshap::explain(
model,
X = xreg,
nsim = 100,
newdata = xreg_pred,
pred_wrapper = function(object, newdata) {
fc_prediction <- forecast::forecast(
object = object,
xreg = newdata,
lambda = NULL,
biasadj = TRUE
)
return(fc_prediction$mean %>% as.numeric())
},
shap_only = FALSE
)
shp <- shapviz::shapviz(shap_vals)
} else {
shp <- NULL
}
list(dt_predictions = dt_predictions, model = model, importance = shp)
}
#' Run random forest model with ranger
#'
#' This function trains a random forest model (ranger) on the
#' specified training dataset and makes predictions on the test dataset in a
#' counterfactual scenario. The model uses meteorological variables and temporal features.
#'
#' Note: Runs the random forest model for individualised use with own data pipeline.
#' Otherwise use [ubair::run_counterfactual()] to call this function.
#' @param train Dataframe of train data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param test Dataframe of test data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param model_params list of hyperparameters to use in ranger call. See \code{\link[ranger:ranger]{ranger:ranger()}} for options.
#' @param alpha Confidence level of the prediction interval between 0 and 1.
#' @param calc_shaps Boolean value. If TRUE, calculate SHAP values for the
#' method used and format them so they can be visualised with \code{\link[shapviz:sv_importance]{shapviz:sv_importance()}} and
#' \code{\link[shapviz:sv_dependence]{shapviz:sv_dependence()}}.
#' The SHAP values are generated for a subset (or all, depending on the size of the dataset) of the
#' test data.
#' @return List with data frame of predictions and model
#' @import ranger
#' @export
run_rf <- function(train, test, model_params, alpha, calc_shaps) {
function_parameters <- list(
y = train$value,
x = train %>% select(-value),
importance = "none",
splitrule = "variance",
keep.inbag = TRUE,
quantreg = TRUE
)
function_parameters <- append(function_parameters, model_params)
model <- do.call(ranger::ranger, args = function_parameters)
quantiles <- data.table::as.data.table(predict(model,
test,
type = "quantiles",
quantiles = c((1 - alpha) / 2, (1 + alpha) / 2)
)$predictions)
if (calc_shaps) {
rlang::check_installed(c("treeshap", "shapviz"),
reason = "calculate shap values for random forest"
)
shap_subset <- test[sample(nrow(test), min(nrow(test), 1000)), ] %>% select(-value)
unified <- treeshap::ranger.unify(model, data.matrix(train %>% select(-value)))
treeshap_vals <- treeshap::treeshap(unified, data.matrix(shap_subset), verbose = FALSE)
shp <- shapviz::shapviz(treeshap_vals, X_pred = data.matrix(shap_subset), X = shap_subset)
} else {
shp <- NULL
}
test[, "prediction" := predict(model, test)$predictions]
test[, c("prediction_lower", "prediction_upper") := quantiles]
list(dt_predictions = test, model = model, importance = shp)
}
#' Run gradient boosting model with lightgbm
#'
#' This function trains a gradient boosting model (lightgbm) on the
#' specified training dataset and makes predictions on the test dataset in a
#' counterfactual scenario. The model uses meteorological variables and temporal features.
#'
#' Note: Runs the gradient boosting model for individualised use with own data pipeline.
#' Otherwise use [ubair::run_counterfactual()] to call this function.
#' @param train Dataframe of train data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param test Dataframe of test data as returned by the [ubair::split_data_counterfactual()]
#' function.
#' @param model_params list of hyperparameters to use in lgb.train call.
#' See \code{\link[lightgbm:lgb.train]{lightgbm:lgb.train()}} params argument for details.
#' @param alpha Confidence level of the prediction interval between 0 and 1.
#' @param calc_shaps Boolean value. If TRUE, calculate SHAP values for the
#' method used and format them so they can be visualised with \code{\link[shapviz:sv_importance]{shapviz:sv_importance()}} and
#' \code{\link[shapviz:sv_dependence]{shapviz:sv_dependence()}}.
#' The SHAP values are generated for a subset (or all, depending on the size of the dataset) of the
#' test data.
#' @return List with data frame of predictions and model
#' @import lightgbm
#' @export
run_lightgbm <- function(train, test, model_params, alpha, calc_shaps) {
dtrain <- lgb.Dataset(
data = data.matrix(train %>% select(-value)),
label = train$value
)
model_mean <- lgb.train(
params = model_params[names(model_params) != "nrounds"],
data = dtrain,
nrounds = model_params$nrounds
)
params_lower <- append(
model_params[names(model_params) != "nrounds"],
list(
"objective" = "quantile",
"alpha" = (1 - alpha) / 2,
"verbosity" = 0
)
)
model_lower <- lgb.train(
params = params_lower,
data = dtrain,
nrounds = model_params$nrounds
)
params_upper <- append(
model_params[names(model_params) != "nrounds"],
list(
"objective" = "quantile",
"alpha" = (1 + alpha) / 2,
"verbosity" = 0
)
)
model_upper <- lgb.train(
params = params_upper,
data = dtrain,
nrounds = model_params$nrounds
)
dapply <- test %>% select(!!colnames(train))
dapply <- data.matrix(dapply %>% select(-value))
dt_predictions <- test
dt_predictions$prediction <- predict(model_mean, dapply)
dt_predictions$prediction_lower <- predict(model_lower, dapply)
dt_predictions$prediction_upper <- predict(model_upper, dapply)
if (calc_shaps) {
rlang::check_installed(c("shapviz"),
reason = "calculate shap values for lightgbm"
)
shp <- shapviz::shapviz(model_mean, X_pred = dapply, X = dapply)
} else {
shp <- NULL
}
list(dt_predictions = dt_predictions, model = model_mean, importance = shp)
}
#' Train a Feedforward Neural Network (FNN) in a Counterfactual Scenario.
#'
#' Trains a feedforward neural network (FNN) model on the
#' specified training dataset and makes predictions on the test dataset in a
#' counterfactual scenario. The model uses meteorological variables and
#' sin/cosine-transformed features. Scales the data before training and rescales
#' predictions, as the model does not converge with unscaled data.
#'
#' This function provides flexibility for users with their own data pipelines
#' or workflows. For a simplified pipeline, consider using
#' \code{\link[ubair:run_counterfactual]{run_counterfactual()}}.
#'
#' Experiment with hyperparameters such as \code{learning_rate},
#' \code{batchsize}, \code{hidden_layers}, and \code{num_epochs} to improve
#' performance.
#'
#' Warning: Using many or large hidden layers in combination with a high number
#' of epochs can lead to long training times.
#'
#' @param train A data frame or tibble containing the training dataset,
#' including the target variable (`value`)
#' and meteorological variables specified in `params$meteo_variables`.
#' @param test A data frame or tibble containing the test dataset on which
#' predictions will be made,
#' using the same meteorological variables as in the training dataset.
#' @param params A list of parameters that define the following:
#' \describe{
#' \item{meteo_variables}{A character vector specifying the names of the
#' meteorological variables used as inputs.}
#' \item{fnn}{A list of hyperparameters for training the feedforward neural
#' network, including:
#' \itemize{
#' \item \code{activation_fun}: The activation function for the hidden
#' layers (e.g., "sigmoid", "tanh").
#' \item \code{momentum}: The momentum factor for training.
#' \item \code{learningrate_scale}: Factor for adjusting learning rate.
#' \item \code{output_fun}: The activation function for the output layer
#' \item \code{batchsize}: The size of the batches during training.
#' \item \code{hidden_dropout}: Dropout rate for the hidden layers to
#' prevent overfitting.
#' \item \code{visible_dropout}: Dropout rate for the input layer.
#' \item \code{hidden_layers}: A vector specifying the number of neurons
#' in each hidden layer.
#' \item \code{num_epochs}: Number of epochs (iterations) for training.
#' \item \code{learning_rate}: Initial learning rate.
#' }
#' }
#' }
#' @param calc_shaps Boolean value. If TRUE, calculate SHAP values for the
#' method used and format them so they can be visualised with
#' \code{\link[shapviz:sv_importance]{shapviz:sv_importance()}} and
#' \code{\link[shapviz:sv_dependence]{shapviz:sv_dependence()}}.
#' The SHAP values are generated for a subset (or all, depending on the size of the dataset) of the
#' test data.
#' @return A list with three elements:
#' \describe{
#' \item{\code{dt_predictions}}{A data frame containing the test data along
#' with the predicted values:
#' \describe{
#' \item{\code{prediction}}{The predicted values from the FNN model.}
#' \item{\code{prediction_lower}}{The same predicted values, as no
#' quantile model is available yet for FNN.}
#' \item{\code{prediction_upper}}{The same predicted values, as no
#' quantile model is available yet for FNN.}
#' }
#' }
#' \item{\code{model}}{The trained FNN model object from the
#' \code{deepnet::nn.train()} function.}
#' \item{\code{importance}}{SHAP importance values (if
#' \code{calc_shaps = TRUE}). Otherwise, `NULL`.}
#' }
#' @export
run_fnn <- function(train, test, params, calc_shaps) {
rlang::check_installed(c("deepnet"), reason = "to run a fnn")
train_transformed <- .transform_input(train, params)
test_transformed <- .transform_input(test, params)
scale_result <- scale_data(
train_data = train_transformed,
apply_data = test_transformed
)
train_matrix <- scale_result$train %>%
dplyr::select(-value) %>%
as.matrix()
test_matrix <- scale_result$apply %>%
dplyr::select(-value) %>%
as.matrix()
target_vector <- scale_result$train$value
fnn_function_parameter <- append(
list(x = train_matrix, y = target_vector),
params$fnn
)
model <- do.call(deepnet::nn.train, args = fnn_function_parameter)
prediction <- deepnet::nn.predict(model, test_matrix)
# no quantil model yet for feedforward neural network
test <- test %>%
mutate(
prediction = prediction,
prediction_lower = prediction,
prediction_upper = prediction
)
test <- rescale_predictions(
scale_result = scale_result,
dt_predictions = test
)
if (calc_shaps) {
rlang::check_installed(c("fastshap", "shapviz"),
reason = "calculate shap values for fnn"
)
shap_vals <- fastshap::explain(
model,
X = train_matrix,
nsim = 50,
newdata = test_matrix,
pred_wrapper = function(object, newdata) {
deepnet::nn.predict(
nn = object,
x = newdata
) %>% as.numeric()
},
shap_only = FALSE
)
shp <- shapviz::shapviz(shap_vals)
} else {
shp <- NULL
}
list(dt_predictions = test, model = model, importance = shp)
}
#' Make fourier features out of hour, weekday and day of the year
#'
#' @return Numeric matrix of sin and cos transformed temporal features of dimension
#' n x 6.
#' @noRd
.make_fourier_features <- function(df) {
df$weekday <- df$weekday %>% as.numeric()
fourier_list <- list(
sin((24 * (df$day_julian) + df$hour) / (24 * 365) * 2 * pi),
cos((24 * (df$day_julian) + df$hour) / (24 * 365) * 2 * pi),
sin((24 * (df$weekday - 1) + df$hour) / (24 * 7) * 2 * pi),
cos((24 * (df$weekday - 1) + df$hour) / (24 * 7) * 2 * pi),
sin(df$hour / 24 * 2 * pi),
cos(df$hour / 24 * 2 * pi)
)
res_matrix <- matrix(unlist(fourier_list), ncol = 6, byrow = FALSE)
colnames(res_matrix) <- c(
"sin_year", "cos_year", "sin_week", "cos_week",
"sin_hour", "cos_hour"
)
res_matrix
}
#' Transform Data to cyclic features for fnn/dynamic regression
#'
#' @return Numeric matrix combining the selected meteorological variables
#' (transformed wind vectors as U and V components) and the Fourier features.
#' @noRd
.transform_input <- function(input_data, params) {
input_meteo <- input_data %>% select(!!params$meteo_variables)
if ("WIG" %in% params$meteo_variables && "WIR" %in% params$meteo_variables) {
input_meteo <- .make_wind_vectors(input_meteo) %>% select(-WIG, -WIR)
}
fourier_features <- .make_fourier_features(input_data)
transformed <- cbind(input_meteo, fourier_features, input_data[, "value"])
transformed
}
#' Creates wind vectors from direction and speed
#'
#' Takes a dataframe with columns WIG (wind speed) and
#' WIR (wind direction in degrees 0 to 360) and creates wind vectors with U and
#' V component
#' @return Data frame of the same format as the input with additional columns
#' "WIND_U" and "WIND_V".
#' @noRd
.make_wind_vectors <- function(dt_prepared) {
dt_prepared <- dt_prepared %>% mutate(
"WIND_U" = sin(pi * WIR / 180) * WIG,
"WIND_V" = cos(pi * WIR / 180) * WIG
)
dt_prepared
}
R/data_cleaning.R 0 → 100644
View file @ 5d21dfb7
#' Clean and Optionally Aggregate Environmental Data
#'
#' Cleans a data table of environmental measurements by filtering for a specific
#' station, removing duplicates, and optionally aggregating the data on a daily
#' basis using the mean.
#'
#' @param env_data A data table in long format.
#' Must include columns:
#' \describe{
#' \item{Station}{Station identifier for the data.}
#' \item{Komponente}{Measured environmental component e.g. temperature, NO2.}
#' \item{Wert}{Measured value.}
#' \item{date}{Timestamp as Date-Time object (`YYYY-MM-DD HH:MM:SS` format).}
#' \item{Komponente_txt}{Textual description of the component.}
#' }
#' @param station Character. Name of the station to filter by.
#' @param aggregate_daily Logical. If `TRUE`, aggregates data to daily mean values. Default is `FALSE`.
#' @return A `data.table`:
#' \itemize{
#' \item If `aggregate_daily = TRUE`: Contains columns for station, component, day, year,
#' and the daily mean value of the measurements.
#' \item If `aggregate_daily = FALSE`: Contains cleaned data with duplicates removed.
#' }
#' @details Duplicate rows (by `date`, `Komponente`, and `Station`) are removed. A warning is issued
#' if duplicates are found.
#' @examples
#' # Example data
#' env_data <- data.table::data.table(
#' Station = c("DENW094", "DENW094", "DENW006", "DENW094"),
#' Komponente = c("NO2", "O3", "NO2", "NO2"),
#' Wert = c(45, 30, 50, 40),
#' date = as.POSIXct(c(
#' "2023-01-01 08:00:00", "2023-01-01 09:00:00",
#' "2023-01-01 08:00:00", "2023-01-02 08:00:00"
#' )),
#' Komponente_txt = c(
#' "Nitrogen Dioxide", "Ozone", "Nitrogen Dioxide", "Nitrogen Dioxide"
#' )
#' )
#'
#' # Clean data for StationA without aggregation
#' cleaned_data <- clean_data(env_data, station = "DENW094", aggregate_daily = FALSE)
#' print(cleaned_data)
#' @export
clean_data <- function(env_data, station, aggregate_daily = FALSE) {
env_data <- .add_year_column(env_data)
env_data <- dplyr::filter(env_data, Station == station)
env_data_unique <- unique(env_data, by = c("date", "Komponente", "Station"))
if (nrow(env_data_unique) < nrow(env_data)) {
warning(sprintf(
"%d duplicate row(s) were removed.",
nrow(env_data) - nrow(env_data_unique)
))
}
if (aggregate_daily) {
env_data_unique <- .aggregate_data(env_data_unique)
}
env_data_unique
}
#' Get Available Meteorological Components
#'
#' Identifies unique meteorological components from the provided environmental data,
#' filtering only those that match the predefined UBA naming conventions. These components
#' include "GLO", "LDR", "RFE", "TMP", "WIG", "WIR", "WIND_U", and "WIND_V".
#' @param env_data Data table containing environmental data.
#' Must contain column "Komponente"
#' @return A vector of available meteorological components.
#' @examples
#' # Example environmental data
#' env_data <- data.table::data.table(
#' Komponente = c("TMP", "NO2", "GLO", "WIR"),
#' Wert = c(25, 40, 300, 50),
#' date = as.POSIXct(c(
#' "2023-01-01 08:00:00", "2023-01-01 09:00:00",
#' "2023-01-01 10:00:00", "2023-01-01 11:00:00"
#' ))
#' )
#' # Get available meteorological components
#' meteo_components <- get_meteo_available(env_data)
#' print(meteo_components)
#' @export
get_meteo_available <- function(env_data) {
meteo_available <- unique(env_data$Komponente) %>%
.[. %in% c("GLO", "LDR", "RFE", "TMP", "WIG", "WIR", "WIND_U", "WIND_V")]
meteo_available
}
#' @return A data.table with an added year column.
#' @noRd
.add_year_column <- function(env_data) {
env_data_copy <- data.table::copy(env_data)
env_data_copy[, year := lubridate::year(date)]
}
#' Adds a `day` column to the data, representing the date truncated to day-level
#' precision. This column is used for later aggregations.
#' @noRd
.add_day_column <- function(env_data) {
env_data %>% dplyr::mutate(day = lubridate::floor_date(date, unit = "day"))
}
#' @return A data.table aggregated to daily mean values and a new column day.
#' @noRd
.aggregate_data <- function(env_data) {
env_data <- .add_day_column(env_data)
env_data[, list(Wert = mean(Wert, na.rm = TRUE)),
by = list(Station, Komponente, Komponente_txt, day, year)
]
}
R/data_loading.R 0 → 100644
View file @ 5d21dfb7
#' Load Parameters from YAML File
#'
#' Reads a YAML file containing model parameters, including station settings,
#' variables, and configurations for various models. If no file path is
#' provided, the function defaults to loading `params.yaml` from the package's
#' `extdata` directory.
#'
#' @param filepath Character. Path to the YAML file. If `NULL`, the function
#' will attempt to load the default `params.yaml` provided in the package.
#' @return A list containing the parameters loaded from the YAML file.
#' @details
#' The YAML file should define parameters in a structured format, such as:
#'
#' ```yaml
#' target: 'NO2'
#'
#' lightgbm:
#' nrounds: 200
#' eta: 0.03
#' num_leaves: 32
#'
#' dynamic_regression:
#' ntrain: 8760
#'
#' random_forest:
#' num.trees: 300
#' max.depth: 10
#'
#' meteo_variables:
#' - GLO
#' - TMP
#' ```
#' @examples
#' \dontrun{
#' params <- load_params("path/to/custom_params.yaml")
#' }
#' @export
#' @importFrom yaml read_yaml
load_params <- function(filepath = NULL) {
if (is.null(filepath)) {
filepath <- system.file("extdata", "params.yaml", package = "ubair")
}
if (file.exists(filepath)) {
params <- yaml::read_yaml(filepath)
return(params)
} else {
stop("YAML file not found at the specified path.")
}
}
#' Copy Default Parameters File
#'
#' Copies the default `params.yaml` file, included with the package, to a
#' specified destination directory. This is useful for initializing parameter
#' files for custom edits.
#'
#' @param dest_dir Character. The path to the directory where the `params.yaml`
#' file will be copied.
#' @return Nothing is returned. A message is displayed upon successful copying.
#' @details
#' The `params.yaml` file contains default model parameters for various
#' configurations such as LightGBM, dynamic regression, and others. See the
#' \code{\link[ubair:load_params]{load_params()}}` documentation for an example of the file's structure.
#'
#' @examples
#' \dontrun{
#' copy_default_params("path/to/destination")
#' }
#' @export
copy_default_params <- function(dest_dir) {
if (!dir.exists(dest_dir)) {
stop("Destination directory does not exist.")
}
file.copy(system.file("extdata", "params.yaml", package = "ubair"),
file.path(dest_dir, "params.yaml"),
overwrite = TRUE
)
message("Default params.yaml copied to ", normalizePath(dest_dir))
}
#' Load UBA Data from Directory
#'
#' This function loads data from CSV files in the specified directory. It supports two formats:
#'
#' 1. "inv": Files must contain the following columns:
#' - `Station`, `Komponente`, `Datum`, `Uhrzeit`, `Wert`.
#' 2. "24Spalten": Files must contain:
#' - `Station`, `Komponente`, `Datum`, and columns `Wert01`, ..., `Wert24`.
#'
#' File names should include "inv" or "24Spalten" to indicate their format. The function scans
#' recursively for `.csv` files in subdirectories and combines the data into a single `data.table`
#' in long format.
#' Files that are not in the exected format will be ignored.
#'
#' @param data_dir Character. Path to the directory containing `.csv` files.
#' @return A `data.table` containing the loaded data in long format. Returns an error if no valid
#' files are found or the resulting dataset is empty.
#' @export
#' @importFrom lubridate ymd_hm ymd_h
#' @importFrom tidyr gather
#' @importFrom dplyr mutate select %>%
load_uba_data_from_dir <- function(data_dir) {
if (!dir.exists(data_dir)) {
stop(paste("Directory does not exist:", data_dir))
}
all_files <- list.files(data_dir,
pattern = "\\.csv$",
recursive = TRUE,
full.names = TRUE
)
list_data_parts <- lapply(unique(dirname(all_files)), function(dir) {
.load_data(dir)
})
combined_data <- data.table::rbindlist(list_data_parts, fill = TRUE)
if (nrow(combined_data) == 0) {
stop(paste(
"The resulting data is empty after loading all files from",
data_dir
))
}
return(combined_data)
}
#' Load data for a Specific Directory
#'
#' @param data_dir Character. Path to the directory containing `.csv` files.
#' @return A `data.table` with the loaded data for the directory. Returns an
#' empty `data.table` if no files match the expected format.
#' @noRd
.load_data <- function(data_dir) {
data_files <- list.files(data_dir, pattern = "\\.csv$")
list_data <- lapply(data_files, function(file) {
.load_data_file(data_dir, file)
})
data.table::rbindlist(list_data, fill = TRUE)
}
#' Load data from a specific file
#'
#' Loads data from a file in "inv" or "24Spalten" format. Unsupported formats
#' return an empty `data.table`.
#'
#' @param data_dir Character. Base directory containing the file.
#' @param file Character. Name of the file to load.
#' @return `data.table` containing loaded data or empty data.table for
#' unsupported formats.
#' @noRd
.load_data_file <- function(data_dir, file) {
file_path <- file.path(data_dir, file)
if (grepl("inv", file)) {
.load_inv_file(file_path, file)
} else if (grepl("24Spalten", file)) {
.load_spalten_file(file_path, file)
} else {
data.table::data.table()
}
}
#' Load data from an 'inv' file
#'
#' @param file_path Full path to the 'inv' file.
#' @param file The filename being loaded.
#' @return A data.table with the loaded 'inv' data.
#' @noRd
.load_inv_file <- function(file_path, file) {
data.table::fread(file_path,
quote = "'",
na.strings = c(
"-999", "-888", "-777", "-666", "555", "-555", "-333",
"-111", "555.00000000"
)
) %>%
mutate(
date = ymd_hm(paste(Datum, Uhrzeit)),
Komponente_txt = Komponente,
Komponente = substring(sub("\\_.*", "", file), 7)
)
}
#' Load data from a '24Spalten' file
#'
#' @param file_path Full path to the 'Spalten' file.
#' @param file The filename being loaded.
#' @return `data.table` containing the processed data from the '24Spalten' file.
#' @noRd
.load_spalten_file <- function(file_path, file) {
data.table::fread(file_path,
quote = "'",
na.strings = c(
"-999", "-888", "-777", "-666", "555", "-555", "-333",
"-111", "555.00000000"
)
) %>%
tidyr::gather("time", "Wert", Wert01:Wert24) %>%
mutate(
Uhrzeit = substring(time, 5),
date = ymd_h(paste(Datum, Uhrzeit)),
Komponente_txt = Komponente,
Komponente = substring(sub("\\_.*", "", file), 8)
) %>%
dplyr::select(-c(Nachweisgrenze, Lieferung))
}
R/data_preprocessing.R 0 → 100644
View file @ 5d21dfb7
#' Removes trend from data
#'
#' Takes a list of train and application data as prepared by
#' [ubair::split_data_counterfactual()]
#' and removes a polynomial, exponential or cubic spline spline trend function.
#' Trend is obtained only from train data. Use as part of preprocessing before
#' training a model based on decision trees, i.e. random forest and lightgbm.
#' For the other methods it may be helpful but they are generally able to
#' deal with trends themselves. Therefore we recommend to try out different
#' versions and guide decisisions using the model evaluation metrics from
#' [ubair::calc_performance_metrics()].
#'
#' Apply [ubair::retrend_predictions()] to predictions to return to the
#' original data units.
#'
#' @param split_data List of two named dataframes called train and apply
#' @param mode String which defines type of trend is present. Options are
#' "linear", "quadratic", "exponential", "spline", "none".
#' "none" returns original data
#' @param num_splines Defines the number of cubic splines if `mode="spline"`.
#' Choose num_splines=1 for cubic polynomial trend. If `mode!="spline"`, this
#' parameter is ignored
#' @param log_transform If `TRUE`, use a log-transformation before detrending
#' to ensure positivity of all predictions in the rest of the pipeline.
#' A exp transformation is necessary during retrending to return to the solution
#' space. Use only in combination with `log_transform` parameter in
#' [ubair::retrend_predictions()]
#' @return List of 3 elements. 2 dataframes: detrended train, apply and the
#' trend function
#' @examples
#' \dontrun{
#' split_data <- split_data_counterfactual(
#' dt_prepared, training_start,
#' training_end, application_start, application_end
#' )
#' detrended_list <- detrend(split_data, mode = "linear")
#' detrended_train <- detrended_list$train
#' detrended_apply <- detrended_list$apply
#' trend <- detrended_list$model
#' }
#' @export
#' @importFrom stats lm predict spline
#' @import splines
detrend <- function(split_data,
mode = "linear",
num_splines = 5,
log_transform = FALSE) {
dt_train_new <- data.table::copy(split_data$train)
dt_apply_new <- data.table::copy(split_data$apply)
stopifnot("log_transform needs to be boolean, i.e. either TRUE or FALSE" = class(log_transform) == "logical")
if (log_transform) {
dt_train_new$value <- log(dt_train_new$value)
dt_apply_new$value <- log(dt_apply_new$value)
}
if (mode == "linear") {
trend <- lm(value ~ date_unix, data = dt_train_new)
} else if (mode == "quadratic") {
trend <- lm(value ~ date_unix + I(date_unix^2), data = dt_train_new)
} else if (mode == "exponential") {
trend <- lm(value ~ log(date_unix), data = dt_train_new)
} else if (mode == "spline") {
stopifnot("Set num_splines larger or equal to 1" = num_splines >= 1)
knots <- seq(
from = min(dt_train_new$date_unix),
to = max(dt_train_new$date_unix),
length.out = num_splines + 1
)
knots <- knots[2:(length(knots) - 1)]
if (length(knots) <= 2) { # If 2 or less knots, just fit cubic polynomial
trend <- lm(value ~ date_unix + I(date_unix^2) + I(date_unix^3),
data = dt_train_new
)
} else { # Else use splines
trend <- lm(value ~ bs(date_unix, knots = knots), data = dt_train_new)
}
} else if (mode == "none") {
trend <- lm(value ~ 1 - 1, data = dt_train_new)
} else {
stop("mode needs to be any of the following strings: 'linear',
'quadratic', 'exponential', 'spline', 'none'")
}
trend_train_values <- predict(trend, newdata = dt_train_new)
dt_train_new$value <- dt_train_new$value - trend_train_values
trend_apply_values <- predict(trend, newdata = dt_apply_new)
dt_apply_new$value <- dt_apply_new$value - trend_apply_values
return(list(train = dt_train_new, apply = dt_apply_new, model = trend))
}
#' Restors the trend in the prediction
#'
#' Takes a dataframe of predictions as returned by any of
#' the 'run_model' functions and restores a trend which was previously
#' removed via [ubair::detrend()]. This is necessary for the predictions
#' and the true values to have the same units. The function is basically
#' the inverse function to [ubair::detrend()] and should only be used in
#' combination with it.
#'
#' @param dt_predictions Dataframe of predictions with columns `value`,
#' `prediction`, `prediction_lower`, `prediction_upper`
#' @param trend lm object generated by [ubair::detrend()]
#' @param log_transform Returns values to solution space, if they have been
#' log transformed during detrending. Use only in combination with `log_transform`
#' parameter in detrend function.
#' @return Retrended dataframe with same structure as `dt_predictions`
#' which is returned by any of the run_model() functions.
#' @examples
#' \dontrun{
#' detrended_list <- detrend(split_data,
#' mode = detrending_function,
#' log_transform = log_transform
#' )
#' trend <- detrended_list$model
#' detrended_train <- detrended_list$train
#' detrended_apply <- detrended_list$apply
#' detrended_train <- detrended_train %>% select(value, dplyr::any_of(variables))
#' result <- run_lightgbm(
#' train = detrended_train,
#' test = detrended_apply,
#' model_params = params$lightgbm,
#' alpha = 0.9,
#' calc_shaps = FALSE
#' )
#' retrended_predictions <- retrend_predictions(result$dt_predictions, trend)
#' }
#' @export
retrend_predictions <- function(dt_predictions, trend, log_transform = FALSE) {
stopifnot("log_transform needs to be boolean, i.e. TRUE or FALSE" = class(log_transform) == "logical")
stopifnot("trend object needs to be a linear model of class 'lm'" = class(trend) == "lm")
stopifnot(
"Not all of 'value', 'prediction', 'prediction_lower', 'prediction_upper' are columns in dt_predictions" =
all(c("value", "prediction", "prediction_lower", "prediction_upper")
%in% colnames(dt_predictions))
)
trend_value <- predict(trend, newdata = dt_predictions)
if (log_transform) {
dt_predictions[,
c("value", "prediction", "prediction_lower", "prediction_upper") := lapply(.SD, \(x) exp(x + trend_value)),
.SDcols = c("value", "prediction", "prediction_lower", "prediction_upper")
]
} else {
dt_predictions[,
c("value", "prediction", "prediction_lower", "prediction_upper") := lapply(.SD, \(x) x + trend_value),
.SDcols = c("value", "prediction", "prediction_lower", "prediction_upper")
]
}
dt_predictions
}
#' Standardize Training and Application Data
#'
#' This function standardizes numeric columns of the `train_data` and applies
#' the same scaling (mean and standard deviation) to the corresponding columns
#' in `apply_data`. It returns the standardized data along with the scaling
#' parameters (means and standard deviations). This is particularly important
#' for neural network approaches as they tend to be numerically unstable and
#' deteriorate otherwise.
#'
#' @param train_data A data frame containing the training dataset to be
#' standardized. It must contain numeric columns.
#' @param apply_data A data frame containing the dataset to which the scaling
#' from `train_data` will be applied.
#'
#' @return A list containing the following elements:
#' \item{train}{The standardized training data.}
#' \item{apply}{The `apply_data` scaled using the means and standard deviations
#' from the `train_data`.}
#' \item{means}{The means of the numeric columns in `train_data`.}
#' \item{sds}{The standard deviations of the numeric columns in `train_data`.}
#' @export
#' @importFrom dplyr mutate across where
#' @examples
#' \dontrun{
#' scale_result <- scale_data(
#' train_data = detrended_list$train,
#' apply_data = detrended_list$apply, scale = TRUE
#' )
#' scaled_train <- scale_result$train
#' scaled_apply <- scale_result$apply
#' }
scale_data <- function(train_data,
apply_data) {
means <- attr(
scale(train_data %>% select(where(is.numeric))),
"scaled:center"
)
sds <- attr(
scale(train_data %>% select(where(is.numeric))),
"scaled:scale"
)
train_data <- train_data %>%
dplyr::mutate_if(is.numeric, ~ as.numeric(scale(.)))
apply_data <- apply_data %>%
mutate(across(
names(sds),
~ as.numeric(scale(.,
center = means[dplyr::cur_column()],
scale = sds[dplyr::cur_column()]
))
))
list(
train = train_data,
apply = apply_data,
means = means,
sds = sds
)
}
#' Rescale predictions to original scale.
#'
#' This function rescales the predicted values (`prediction`, `prediction_lower`,
#' `prediction_upper`). The scaling is reversed using the means and
#' standard deviations that were saved from the training data. It is the inverse
#' function to [ubair::scale_data()] and should be used only in combination.
#'
#' @param scale_result A list object returned by [ubair::scale_data()],
#' containing the means and standard deviations used for scaling.
#' @param dt_predictions A data frame containing the predictions,
#' including columns `prediction`, `prediction_lower`, `prediction_upper`.
#'
#' @return A data frame with the predictions and numeric columns rescaled back
#' to their original scale.
#'
#' @export
#' @examples
#' \dontrun{
#' scale_res <- scale_data(train_data = train, apply_data = apply)
#' res <- run_fnn(train = scale_res$train, test = scale_res$apply, params)
#' dt_predictions <- res$dt_predictions
#' rescaled_predictions <- rescale_predictions(scale_res, dt_predictions)
#' }
rescale_predictions <- function(scale_result, dt_predictions) {
means <- scale_result$means
sds <- scale_result$sds
rescaled_predictions <- dt_predictions %>%
mutate(
prediction = prediction * sds["value"] + means["value"],
prediction_lower = prediction_lower * sds["value"] + means["value"],
prediction_upper = prediction_upper * sds["value"] + means["value"]
)
return(rescaled_predictions)
}
R/model_evaluation.R 0 → 100644
View file @ 5d21dfb7
#' Calculates performance metrics of a business-as-usual model
#'
#' Model agnostic function to calculate a number of common performance
#' metrics on the reference time window.
#' Uses the true data `value` and the predictions `prediction` for this calculation.
#' The coverage is calculated from the columns `value`, `prediction_lower` and
#' `prediction_upper`.
#' Removes dates in the effect and buffer range as the model is not expected to
#' be performing correctly for these times. The incorrectness is precisely
#' what we are using for estimating the effect.
#' @param predictions data.table or data.frame with the following columns
#' \describe{
#' \item{date}{Date of the observation. Needs to be comparable to
#' date_effect_start element.}
#' \item{value}{True observed value of the station}
#' \item{prediction}{Predicted model output for the same time and station
#' as value}
#' \item{prediction_lower}{Lower end of the prediction interval}
#' \item{prediction_upper}{Upper end of the prediction interval}
#' }
#'
#' @param date_effect_start A date. Start date of the
#' effect that is to be evaluated. The data from this point onwards is disregarded
#' for calculating model performance
#' @param buffer Integer. An additional buffer window before date_effect_start to account
#' for uncertainty in the effect start point. Disregards additional buffer data
#' points for model evaluation
#' @return Named vector with performance metrics of the model
#' @export
calc_performance_metrics <- function(predictions, date_effect_start = NULL, buffer = 0) {
df <- data.table::copy(predictions)
stopifnot("Buffer needs to be larger or equal to 0" = buffer >= 0)
if (!is.null(date_effect_start)) {
stopifnot(
"date_effect_start needs to be NULL or a date object" =
inherits(date_effect_start, "Date") | lubridate::is.POSIXt(date_effect_start)
)
df <- df[date < (date_effect_start - as.difftime(buffer, units = "hours"))]
}
metrics <- c(
"RMSE" = sqrt(mean((df$value - df$prediction)**2)),
"MSE" = mean((df$value - df$prediction)**2),
"MAE" = mean(abs(df$value - df$prediction)),
"MAPE" = mean(abs(df$value - df$prediction) / (df$value + 1)),
"Bias" = mean(df$prediction - df$value),
"R2" = 1 - sum((df$value - df$prediction)**2) / sum((df$value - mean(df$value))**2),
"Coverage lower" = mean(df$value >= df$prediction_lower),
"Coverage upper" = mean(df$value <= df$prediction_upper),
"Coverage" = mean(df$value >= df$prediction_lower) +
mean(df$value <= df$prediction_upper) - 1,
"Correlation" = stats::cor(df$value, df$prediction),
"MFB" = 2 * mean((df$prediction - df$value) / (df$prediction + df$value)),
"FGE" = 2 * mean(abs(df$prediction - df$value) / (df$prediction + df$value))
)
metrics
}
#' Calculates summary statistics for predictions and true values
#'
#' Helps with analyzing predictions by comparing them with the true values on
#' a number of relevant summary statistics.
#' @param predictions Data.table or data.frame with the following columns
#' \describe{
#' \item{date}{Date of the observation. Needs to be comparable to
#' date_effect_start element.}
#' \item{value}{True observed value of the station}
#' \item{prediction}{Predicted model output for the same time and station
#' as value}
#' }
#' @param date_effect_start A date. Start date of the
#' effect that is to be evaluated. The data from this point onwards is disregarded
#' for calculating model performance
#' @param buffer Integer. An additional buffer window before date_effect_start to account
#' for uncertainty in the effect start point. Disregards additional buffer data
#' points for model evaluation
#' @return data.frame of summary statistics with columns true and prediction
#' @export
calc_summary_statistics <- function(predictions, date_effect_start = NULL, buffer = 0) {
df <- data.table::copy(predictions)
stopifnot("Buffer needs to be larger or equal to 0" = buffer >= 0)
if (!is.null(date_effect_start)) {
stopifnot(
"date_effect_start needs to be NULL or a date object" =
inherits(date_effect_start, "Date") | lubridate::is.POSIXt(date_effect_start)
)
df <- df[date < (date_effect_start - as.difftime(buffer, units = "hours"))]
}
data.frame(
true = c(
min(df$value),
max(df$value),
stats::var(df$value),
mean(df$value),
stats::quantile(df$value, probs = 0.05),
stats::quantile(df$value, probs = 0.25),
stats::median(df$value),
stats::quantile(df$value, probs = 0.75),
stats::quantile(df$value, probs = 0.95)
),
prediction = c(
min(df$prediction),
max(df$prediction),
stats::var(df$prediction),
mean(df$prediction),
stats::quantile(df$prediction, probs = 0.05),
stats::quantile(df$prediction, probs = 0.25),
stats::median(df$prediction),
stats::quantile(df$prediction, probs = 0.75),
stats::quantile(df$prediction, probs = 0.95)
),
row.names = c(
"min", "max", "var", "mean", "5-percentile",
"25-percentile", "median/50-percentile",
"75-percentile", "95-percentile"
)
)
}
#' Estimates size of the external effect
#'
#' Calculates an estimate for the absolute and relative effect size of the
#' external effect. The absolute effect is the difference between the model
#' bias in the reference time and the effect time windows. The relative effect
#' is the absolute effect divided by the mean true value in the reference
#' window.
#'
#' Note: Since the bias is of the model is an average over predictions and true
#' values, it is important, that the effect window is specified correctly.
#' Imagine a scenario like a fire which strongly affects the outcome for one
#' hour and is gone the next hour. If we use a two week effect window, the
#' estimated effect will be 14*24=336 times smaller compared to using a 1-hour
#' effect window. Generally, we advise against studying very short effects (single
#' hour or single day). The variability of results will be too large to learn
#' anything meaningful.
#'
#' @param df Data.table or data.frame with the following columns
#' \describe{
#' \item{date}{Date of the observation. Needs to be comparable to
#' date_effect_start element.}
#' \item{value}{True observed value of the station}
#' \item{prediction}{Predicted model output for the same time and station
#' as value}
#' }
#' @param date_effect_start A date. Start date of the
#' effect that is to be evaluated. The data from this point onward is disregarded
#' for calculating model performance.
#' @param buffer Integer. An additional buffer window before date_effect_start to account
#' for uncertainty in the effect start point. Disregards additional buffer data
#' points for model evaluation
#' @param verbose Prints an explanation of the results if TRUE
#' @return A list with two numbers: Absolute and relative estimated effect size.
#' @export
estimate_effect_size <- function(df, date_effect_start, buffer = 0, verbose = FALSE) {
stopifnot(
"date_effect_start needs to be NULL or a date object" =
inherits(date_effect_start, "Date") | lubridate::is.POSIXt(date_effect_start)
)
stopifnot("Buffer needs to be larger or equal to 0" = buffer >= 0)
reference <- df[date < (date_effect_start - as.difftime(buffer, units = "hours"))]
effect <- df[date >= date_effect_start]
bias_ref <- mean(reference$prediction - reference$value)
bias_effect <- mean(effect$prediction - effect$value)
effectsize <- bias_ref - bias_effect
rel_effectsize <- effectsize / mean(effect$value)
rel_effectsize <- round(rel_effectsize, 4)
if (verbose) {
cat(sprintf("The external effect changed the target value on average by %.3f compared to the reference time window. This is a %1.2f%% relative change.", effectsize, 100 * rel_effectsize))
}
list(absolute_effect = effectsize, relative_effect = rel_effectsize)
}
R/modelling.R 0 → 100644
View file @ 5d21dfb7
#' Prepare Data for Training a model
#'
#' Prepares environmental data by filtering for relevant components,
#' converting the data to a wide format, and adding temporal features. Should be
#' called before
#' \code{\link[ubair:split_data_counterfactual]{split_data_counterfactual()}}
#'
#' @param env_data A data table in long format.
#' Must include the following columns:
#' \describe{
#' \item{Station}{Station identifier for the data.}
#' \item{Komponente}{The environmental component being measured
#' (e.g., temperature, NO2).}
#' \item{Wert}{The measured value of the component.}
#' \item{date}{Timestamp as `POSIXct` object in `YYYY-MM-DD HH:MM:SS` format.}
#' \item{Komponente_txt}{A textual description of the component.}
#' }
#' @param params A list of modelling parameters loaded from `params.yaml`.
#' Must include:
#' \describe{
#' \item{meteo_variables}{A vector of meteorological variable names.}
#' \item{target}{The name of the target variable.}
#' }
#' @return A `data.table` in wide format, with columns:
#' `date`, one column per component, and temporal features
#' like `date_unix`, `day_julian`, `weekday`, and `hour`.
#' @examples
#' env_data <- data.table::data.table(
#' Station = c("StationA", "StationA", "StationA"),
#' Komponente = c("NO2", "TMP", "NO2"),
#' Wert = c(50, 20, 40),
#' date = as.POSIXct(c("2023-01-01 10:00:00", "2023-01-01 11:00:00", "2023-01-02 12:00:00"))
#' )
#' params <- list(meteo_variables = c("TMP"), target = "NO2")
#' prepared_data <- prepare_data_for_modelling(env_data, params)
#' print(prepared_data)
#'
#' @export
prepare_data_for_modelling <- function(env_data, params) {
components <- c(params$meteo_variables, params$target)
dt_filtered <- .extract_components(env_data, components)
dt_wide <- .cast_to_wide(dt_filtered)
if (!params$target %in% names(dt_wide)) {
warning(sprintf("Target '%s' is not present in the data. Make sure it exists
and you have set the correct target name", params$target))
stop("Exiting function due to missing target data.")
}
dt_prepared <- dt_wide %>%
dplyr::rename(value = params$target) %>%
.add_date_variables(replace = TRUE) %>%
dplyr::filter(!is.na(value))
dt_prepared
}
#' Turn date feature into temporal features date_unix, day_julian, weekday and
#' hour
#'
#' @param df Data.table with column date formatted as date-time object
#' @param replace Boolean which determines whether to replace existing temporal variables
#' @return A data.table with all relevant temporal features for modelling
#' @noRd
.add_date_variables <- function(df, replace) {
names <- names(df)
if (replace) {
df$date_unix <- as.numeric(df$date)
df$day_julian <- lubridate::yday(df$date)
df$weekday <- .wday_monday(df$date, as.factor = TRUE)
df$hour <- lubridate::hour(df$date)
} else {
if (!"date_unix" %in% names) {
df$date_unix <- as.numeric(df$date)
}
if (!"day_julian" %in% names) {
df$day_julian <- lubridate::yday(df$date)
}
if (!"weekday" %in% names) {
df$weekday <- .wday_monday(df$date, as.factor = TRUE)
}
if (!"hour" %in% names) {
df$hour <- lubridate::hour(df$date)
}
}
return(df)
}
#' Reformat lubridate weekdays into weekdays with monday as day 1
#'
#' @param x Vector of date-time objects
#' @param as.factor Boolean that determines whether to return a factor or a numeric vector
#' @noRd
.wday_monday <- function(x, as.factor = FALSE) {
x <- lubridate::wday(x)
x <- x - 1
x <- ifelse(x == 0, 7, x)
if (as.factor) {
x <- factor(x, levels = 1:7, ordered = TRUE)
}
return(x)
}
#' Split Data into Training and Application Datasets
#'
#' Splits prepared data into training and application datasets based on
#' specified date ranges for a business-as-usual scenario. Data before
#' `application_start` and after `application_end` is used as training data,
#' while data within the date range is used for application.
#'
#' @param dt_prepared The prepared data table.
#' @param application_start The start date(date object) for the application
#' period of the business-as-usual simulation. This coincides with the start of
#' the reference window.
#' Can be created by e.g. lubridate::ymd("20191201")
#' @param application_end The end date(date object) for the application period
#' of the business-as-usual simulation. This coincides with the end of
#' the effect window.
#' Can be created by e.g. lubridate::ymd("20191201")
#' @return A list with two elements:
#' \describe{
#' \item{train}{Data outside the application period.}
#' \item{apply}{Data within the application period.}
#' }
#' @examples
#' dt_prepared <- data.table::data.table(
#' date = as.Date(c("2023-01-01", "2023-01-05", "2023-01-10")),
#' value = c(50, 60, 70)
#' )
#' result <- split_data_counterfactual(
#' dt_prepared,
#' application_start = as.Date("2023-01-03"),
#' application_end = as.Date("2023-01-08")
#' )
#' print(result$train)
#' print(result$apply)
#' @export
split_data_counterfactual <- function(dt_prepared,
application_start,
application_end) {
stopifnot(
inherits(application_start, "Date"),
inherits(application_end, "Date")
)
stopifnot(application_start <= application_end)
dt_train <- dt_prepared[date < application_start | date > application_end]
dt_apply <- dt_prepared[date >= application_start & date <= application_end]
list(train = dt_train, apply = dt_apply)
}
#' Extract Components for Modelling
#' Stop with error message if any selected meteo variable/component is not
#' contained in the data.
#'
#' @param env_data Daily aggregated data table.
#' @param components Vector of component names to extract.
#' @return A data.table filtered by the specified components.
#' @noRd
.extract_components <- function(env_data, components) {
if (!all(components %in% env_data$Komponente)) {
missing_components <- components[!components %in% env_data$Komponente]
stop(paste(
"Data does not contain all selected variables:", missing_components,
"\n Check data and meteo_variables/params.yaml."
))
}
env_data[Komponente %in% components, list(Komponente, Wert, date)]
}
#' @param dt_filtered Filtered data.table.
#' @return A wide-format data.table.
#' @noRd
#' @examples
#' dt_filtered <- data.table::data.table(
#' date = as.POSIXct(c("2023-01-01", "2023-01-01", "2023-01-02")),
#' Komponente = c("NO2", "TMP", "NO2"),
#' Wert = c(50, 20, 40)
#' )
#' wide_data <- .cast_to_wide(dt_filtered)
#' print(wide_data)
.cast_to_wide <- function(dt_filtered) {
data.table::dcast(dt_filtered,
formula = date ~ Komponente,
value.var = "Wert"
)
}
R/sample_data_DESN025.R 0 → 100644
View file @ 5d21dfb7
#' Environmental Data for Modelling from station DESN025 in Leipzig-Mitte.
#'
#' A dataset containing environmental measurements collected at station in
#' Leipzig Mitte with observations of different environmental components over
#' time. This data is used for environmental modelling tasks, including
#' meteorological variables and other targets.
#'
#' @format ## sample_data_DESN025
#' A data table with the following columns:
#' \describe{
#' \item{Station}{Station identifier where the data was collected.}
#' \item{Komponente}{The environmental component being measured
#' (e.g., temperature, NO2).}
#' \item{Wert}{The measured value of the component.}
#' \item{date}{The timestamp for the observation, formatted as a Date-Time
#' object in the format
#' \code{"YYYY-MM-DD HH:MM:SS"} (e.g., "2010-01-01 07:00:00").}
#' \item{Komponente_txt}{A textual description or label for the component.}
#' }
#'
#' The dataset is structured in a long format and is prepared for further
#' transformation into a wide format for modelling.
#'
#' @source Umweltbundesamt
#' @examples
#' \dontrun{
#' params <- load_params("path/to/params.yaml")
#' dt_prepared <- prepare_data_for_modelling(sample_data_DESN025, params)
#' }
"sample_data_DESN025"
R/utils.R 0 → 100644
View file @ 5d21dfb7
# required to suppress devtools::check() notes that occur from data.table syntax
utils::globalVariables(c(
".", "Datum", "Komponente", "Komponente_txt",
"Lieferung", "Nachweisgrenze", "Station", "Uhrzeit",
"Wert", "Wert01", "Wert24", "day", "part", "prediction_upper",
"prediction_lower", "prediction",
"se", "time", "value", "variable", "WIR", "WIG", "year", "Werte_aggregiert"
))
R/visualisation.R 0 → 100644
View file @ 5d21dfb7
#' Descriptive plot of daily time series data
#'
#' This function produces descriptive time-series plots with smoothing
#' for the meteorological and potential target variables that were measured at a station.
#'
#' @param env_data A data table of measurements of one air quality measurement station.
#' The data should contain the following columns:
#' \describe{
#' \item{Station}{Station identifier where the data was collected.}
#' \item{Komponente}{The environmental component being measured
#' (e.g., temperature, NO2).}
#' \item{Wert}{The measured value of the component.}
#' \item{date}{The timestamp for the observation,
#' formatted as a Date-Time object in the format
#' \code{"YYYY-MM-DD HH:MM:SS"} (e.g., "2010-01-01 07:00:00").}
#' \item{Komponente_txt}{A textual description or label for the component.}
#' }
#' @param variables list of variables to plot. Must be in `env_data$Komponente`.
#' Meteorological variables can be obtained from params.yaml.
#' @param years Optional. A numeric vector, list, or a range specifying the
#' years to restrict the plotted data.
#' You can provide:
#' - A single year: `years = 2020`
#' - A numeric vector of years: `years = c(2019, 2020, 2021)`
#' - A range of years: `years = 2019:2021`
#' If not provided, data for all available years will be used.
#' @param smoothing_factor A number that defines the magnitude of smoothing.
#' Default is 1. Smaller numbers correspond to less smoothing, larger numbers to more.
#' @export
#' @importFrom ggplot2 ggplot aes geom_line facet_wrap geom_smooth
plot_station_measurements <- function(env_data, variables, years = NULL, smoothing_factor = 1) {
stopifnot("No data in the env_data data.table" = nrow(env_data) > 0)
stopifnot(
"More than one station in env_data. Use clean_data to specify which one to use" =
length(unique(env_data$Station)) == 1
)
if (is.null(years)) {
years <- unique(env_data$year)
}
if (!"day" %in% colnames(env_data)) {
env_data <- .aggregate_data(env_data)
}
env_data[, Werte_aggregiert := data.table::frollmean(Wert,
12 * 7 * smoothing_factor,
na.rm = TRUE,
align = "center"
),
by = Komponente_txt
]
p <- ggplot(env_data[Komponente %in% variables & year %in% years], aes(day, Wert)) +
geom_line() +
facet_wrap(~Komponente_txt, scales = "free_y", ncol = 1) +
geom_line(aes(day, Werte_aggregiert), color = "blue", size = 1.5)
p
}
#' Prepare Plot Data and Plot Counterfactuals
#'
#' Smooths the predictions using a rolling mean, prepares the data for plotting,
#' and generates the counterfactual plot for the application window. Data before
#' the red box are reference window, red box is buffer and values after black,
#' dotted line are effect window.
#'
#' The optional grey ribbon is a prediction interval for the hourly values. The
#' interpretation for a 90% prediction interval (to be defined in `alpha` parameter
#' of [ubair::run_counterfactual()]) is that 90% of the true hourly values
#' (not the rolled means) lie within the grey band. This might be helpful for
#' getting an idea of the variance of the data and predictions.
#'
#' @param predictions The data.table containing the predictions (hourly)
#' @param params Parameters for plotting, including the target variable.
#' @param window_size The window size for the rolling mean (default is 14 days).
#' @param date_effect_start A date. Start date of the
#' effect that is to be evaluated. The data from this point onwards is disregarded
#' for calculating model performance
#' @param buffer Integer. An additional, optional buffer window before
#' `date_effect_start` to account for uncertainty in the effect start point.
#' Disregards additional buffer data points for model evaluation.
#' Use `buffer=0` for no buffer.
#' @param plot_pred_interval Boolean. If `TRUE`, shows a grey band of the prediction
#' interval.
#' @return A ggplot object with the counterfactual plot. Can be adjusted further,
#' e.g. set limits for the y-axis for better visualisation.
#' @export
#' @importFrom ggplot2 ggplot aes geom_ribbon geom_line geom_vline theme_bw labs
#' @importFrom data.table melt
#' @importFrom data.table .SD
#' @importFrom data.table :=
plot_counterfactual <- function(predictions, params, window_size = 14,
date_effect_start = NULL, buffer = 0,
plot_pred_interval = TRUE) {
stopifnot("No data in predictions" = nrow(predictions) > 0)
stopifnot(
"Not all of 'value', 'prediction', 'prediction_lower', 'prediction_upper' are present in predictions data.table" =
c("value", "prediction", "prediction_lower", "prediction_upper") %in% colnames(predictions)
)
# Smooth the data using a rolling mean
dt_plot <- data.table::copy(predictions)
dt_plot <- dt_plot[,
c("value", "prediction", "prediction_lower", "prediction_upper") :=
lapply(.SD, \(x) data.table::frollmean(x, 24 * window_size, align = "center")),
.SDcols = c("value", "prediction", "prediction_lower", "prediction_upper")
]
dt_plot <- dt_plot[!is.na(value)]
# Melt the data for plotting
dt_plot_melted <- melt(dt_plot,
id.vars = c("date", "prediction_lower", "prediction_upper"),
measure.vars = c("value", "prediction"),
variable.name = "variable",
value.name = "value"
)
# Prepare the ggplot object
p <- ggplot(dt_plot_melted, aes(x = date))
if (plot_pred_interval) {
p <- p +
geom_ribbon(aes(ymin = prediction_lower, ymax = prediction_upper),
fill = "grey70", alpha = 0.8
)
}
p <- p + geom_line(aes(y = value, color = variable)) +
theme_bw() +
ggplot2::scale_x_datetime(
date_minor_breaks = "1 month",
date_breaks = "2 month"
) +
labs(y = paste(params$target, paste("concentration", window_size, "d rolling mean")))
# Add vertical lines for external effect if provided
if (!is.null(date_effect_start) && length(date_effect_start) == 1) {
p <- p + geom_vline(
xintercept = date_effect_start, linetype = 4,
colour = "black"
)
}
if (buffer > 0) {
p <- p + ggplot2::annotate("rect",
xmin = date_effect_start - as.difftime(buffer, units = "hours"),
xmax = date_effect_start,
ymin = -Inf,
ymax = Inf,
alpha = 0.1,
fill = "red"
)
}
p
}
README.Rmd 0 → 100644
View file @ 5d21dfb7
---
output:
github_document:
df_print: kable
editor_options:
chunk_output_type: console
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r setup, include=FALSE}
knitr::opts_chunk$set(
message = FALSE,
warning = FALSE,
eval = FALSE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
# ubair <img src="inst/sticker/stickers-ubair-1.png" align="right" width="20%"/>
**ubair** is an R package for Statistical Investigation of the Impact of External Conditions on Air Quality: it uses the statistical software R to analyze and visualize the impact of external factors, such as traffic restrictions, hazards, and political measures, on air quality. It aims to provide experts with a transparent comparison of modeling approaches and to support data-driven evaluations for policy advisory purposes.
## Installation
Install via cran or if you have access to
[https://gitlab.ai-env.de/use-case-luft/ubair](https://gitlab.ai-env.de/use-case-luft/ubair)
you can use one of the following options:
#### Using an archive file
Recommended if you do not have git installed.
- Download zip/tar.gz from GitLab
- Start a new R-Project or open an existing one
- in R-Studio:
- go to 'Packages'-Tab (next to Help/Plots/Files)
- Click on 'Install' (left upper corner)
- Install from: choose "Package Archive File"
- Browse to zip-file
- 'Install'
- alternatively, type in console:
```{r install_package_local}
install.packages("<path-to-zip>/ubair-master.zip", repos = NULL, type = "source")
```
#### Using remote package
Git needs to be installed.
```{r install_package_remote}
install.packages("remotes")
# requires a configures ssh-key
remotes::install_git("git@gitlab.ai-env.de:use-case-luft/ubair.git")
# alternative via password
remotes::install_git("https://gitlab.ai-env.de/use-case-luft/ubair.git")
```
## Sample Usage of package
For a more detailed explanation of the package, you can access the vignettes:
- View user_sample source code directly in the [vignettes/](vignettes/) folder.
- Open vignette by function `vignette("user_sample_1", package = "ubair")`,
if the package was installed with vignettes
``` {r load data, eval=TRUE}
library(ubair)
params <- load_params()
env_data <- sample_data_DESN025
```
```{r plot-meteo-data, eval=TRUE, fig.height=10, fig.width=10}
# Plot meteo data
plot_station_measurements(env_data, params$meteo_variables)
```
- split data into training, reference and effect time intervals
<img src="man/figures/time_split_overview.png" width="100%"/>
``` {r counterfactual-scenario, eval=TRUE, fig.height=5, fig.width=10}
application_start <- lubridate::ymd("20191201") # This coincides with the start of the reference window
date_effect_start <- lubridate::ymd_hm("20200323 00:00") # This splits the forecast into reference and effect
application_end <- lubridate::ymd("20200504") # This coincides with the end of the effect window
buffer <- 24 * 14 # 14 days buffer
dt_prepared <- prepare_data_for_modelling(env_data, params)
dt_prepared <- dt_prepared[complete.cases(dt_prepared)]
split_data <- split_data_counterfactual(
dt_prepared, application_start,
application_end
)
res <- run_counterfactual(split_data,
params,
detrending_function = "linear",
model_type = "lightgbm",
alpha = 0.9,
log_transform = TRUE,
calc_shaps = TRUE
)
predictions <- res$prediction
plot_counterfactual(predictions, params,
window_size = 14,
date_effect_start,
buffer = buffer,
plot_pred_interval = TRUE
)
```
```{r evaluation metrics, eval=TRUE}
round(calc_performance_metrics(predictions, date_effect_start, buffer = buffer), 2)
round(calc_summary_statistics(predictions, date_effect_start, buffer = buffer), 2)
estimate_effect_size(predictions, date_effect_start, buffer = buffer, verbose = TRUE)
```
### SHAP feature importances
```{r feature_importance, eval=TRUE, fig.height=4, fig.width=8}
shapviz::sv_importance(res$importance, kind = "bee")
xvars <- c("TMP", "WIG", "GLO", "WIR")
shapviz::sv_dependence(res$importance, v = xvars)
```
## Development
### Prerequisites
1. **R**: Make sure you have R installed (recommended version 4.4.1). You can download it from [CRAN](https://cran.r-project.org/).
2. **RStudio** (optional but recommended): Download from [RStudio](https://www.rstudio.com/).
### Setting Up the Environment
Install the development version of ubair:
```{r}
install.packages("renv")
renv::restore()
devtools::build()
devtools::load_all()
```
### Development
#### Install pre-commit hook (required to ensure tidyverse code formatting)
```
pip install pre-commit
```
#### Add new requirements
If you add new dependencies to *ubair* package, make sure to update the renv.lock file:
``` r
renv::snapshot()
```
#### style and documentation
Before you commit your changes update documentation, ensure style complies with tidyverse styleguide and all tests run without error
```{r}
# update documentation and check package integrity
devtools::check()
# apply tidyverse style (also applied as precommit hook)
usethis::use_tidy_style()
# you can check for existing lintr warnings by
devtools::lint()
# run tests
devtools::test()
# build README.md if any changes have been made to README.Rmd
devtools::build_readme()
```
#### Pre-commit hook
in .pre-commit-hook.yaml pre-commit rules are defined and applied before each commmit.
This includes:
split
- run styler to format code in tidyverse style
- run roxygen to update doc
- check if readme is up to date
- run lintr to finally check code style format
If precommit fails, check the automatically applied changes, stage them and retry to commit.
#### Test Coverage
Install covr to run this.
```{r test_coverage, message=FALSE, warning=FALSE}
cov <- covr::package_coverage(type = "all")
cov_list <- covr::coverage_to_list(cov)
data.table::data.table(
part = c("Total", names(cov_list$filecoverage)),
coverage = c(cov_list$totalcoverage, as.vector(cov_list$filecoverage))
)
```
```{r test_coverage_report}
covr::report(cov)
```
## Contacts
**Jore Noa Averbeck**
[JoreNoa.Averbeck@uba.de](mailto:JoreNoa.Averbeck@uba.de)
**Raphael Franke**
[Raphael.Franke@uba.de](mailto:Raphael.Franke@uba.de)
**Imke Voß**
[imke.voss@uba.de](mailto:imke.voss@uba.de)

Consent

On this website, we use the web analytics service Matomo to analyze and review the use of our website. Through the collected statistics, we can improve our offerings and make them more appealing for you. Here, you can decide whether to allow us to process your data and set corresponding cookies for these purposes, in addition to technically necessary cookies. Further information on data protection—especially regarding "cookies" and "Matomo"—can be found in our privacy policy. You can withdraw your consent at any time.