Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.

Prerequisites

To run this project you need to have python 3.8 or higher installed on your machine.

First, clone this Repository and initialize the submodules:

git clone --recurse-submodule https://git.rwth-aachen.de/acs/public/automation/plf/proloaf.git

or if you have already cloned the project and you are missing e.g. the open data directory for execution run:

git submodule update --init --recursive

(Optional) Create a virtual environment and activate it:

python -m venv venv
source venv/bin/activate

To install all required packages first change into the directory then install the package:

cd proloaf
pip install .

There are two variants with extra dependencies. To install jupyter for being able to use the tutorial notebooks use

pip install .[tutorial]

or use

cd proloaf
pip install -e .[dev]

to install in editable mode and include optional development dependencies.

On low RAM machines the option

pip install . --no-cache-dir

might be necessary.

ProLoaF supports Python 3.8 and higher.

Running the code

This project contains 4 scripts that can be used in conjunction with one another or separately. Configuration for these scripts is given in a config.json file in the subfolders of the /targets/ folder, the config in targets/opsd should be a good starting point.

Example Workflow

1. Add the data a model should be trained on to the /data/ folder

2. Preprocess the data using prep.py or custom scripts if needed. Data including targets should be all in one .csv file separated with ;.

3. Create a new folder in the targets/ folder corresponding to your forecast. Copy the config files from targets/opsd and adjust it to your needs. In the very least everything to do with your data needs to be adjusted (like column names etc). For more information have a look in the descripiton of the config.

4. Train a model using python3 ./src/train.py -s <name_of_folder_containing_configs>. You can also give a path to the config relative to the project folder using -c instead of -s

5. Evaluate the model using ``python3 ./src/evaluate.py.py -s <name_of_folder_containing_configs>` in the same fashion.

6. To understand how to generate a single prediction have a look at the examples.