schneider-datathon

Schneider Electric European Hackathon 2022 datathon dedicated repository

Results

We achieved the 4th place out of 203 teams.

Code execution

In order to launch the code for the predictive model, you have to type:

python -m fpds.run

Bear in mind that if the code is running locally, the LOCAL global variables in the modules have to keep as True; while in case it is running at the cloud you have to adjust the path.

It also assumes that data exists with the default structure, that is all the downloaded folders and csv have to go into data.

Other relevant links

In the following links there are interesting resources:

• EfficientNet notebook
• Transformer (ViT) notebook
• Models' serialization

Local environment

conda create -n schneider python=3.9
conda activate schneider
conda install -c anaconda tensorflow -y
conda install -c conda-forge keras matplotlib transformers -y
pip install vit-keras
conda install -c anaconda seaborn pillow -y
conda install -c esri tensorflow-addons -y
conda install -c anaconda sphinx numpydoc \
    sphinx_rtd_theme recommonmark python-graphviz -y
pip install --upgrade myst-parser

Documentation

Whenever the modules have been updated, the documentation can be re-generated from the docs folder by typing ():

(<condaenv>) schneider-datathon/docs $ make html

Then, opening the schneider-datathon/docs/build/html/index.html file in the browser will display the generated documentation.