ThermoEduTech

cs7643-a2 Summer semester DL project

Google Drive: https://drive.google.com/drive/u/0/folders/1uIVly1sJD1T2tP-W8WJLExcftLjqPXqk

Progress Tracker & Experiments: https://docs.google.com/spreadsheets/d/1uDEg6ZnbloQdvZVw1xnJkmZJi-oku_5_uSUEC6PwWvQ/edit?gid=626672196#gid=626672196

Datasets

1. School Sensor Measurements: Hour, Minute, Seconds resolution
2. SAURAN Solar Radiation Measurements: Hour, Minute resolution

Setup

• Install the required packages pip install -r requirements.txt (NOTE: this will install the torch for the CPU usage)
• Setup the python paths

  chmod +x setup.sh
  ./setup.sh
  

Project structure

This project has this directory structure

• data - contains the necessary dataset csv file
• utils - simple utility and helper functions. It also has additional DataWrapper and Visualizer classes for preprocessing and EDA
• notebooks - simple notebooks for the usage and results reproducibility
• src - source code directory (check there for details)

Usage

Train

In order to train the models, which for this project 3 types are used, Baseline(sklearn, xgboost), RNN/Transformers(torch) with various configurations can be done either by looking at the notebooks directory or in the src/train.py.

• Baseline Models

  ./notebooks/baseline.ipynb
  

• RNN Models

   ./notebooks/rnn.ipynb
  

• GRU and CNN Models

   ./notebooks/CNN_GRU.ipynb
  

• LSTM Models

  ./notebooks/LSTM.ipynb
  

• Transformers

  python src/train.py -c configs/transformer_configs.yaml
  

Evaluate / Test

python src/evaluate.py -c path/to/train/resulting/directory/configs.yaml

Also check the notebooks/Results.ipynb where the comparison of each model is provided.