ecommute berlin - ML & backend app

This is the repository for the ML & backend app of the ecommute berlin project for the data4good hackathon 2024.

Technologies used:

• pytorch
• numpy & pandas
• fastapi
• onnx runtime

Checkout our frontend repo at: https://github.com/Warsaw-Student-Hacking-Team/data4good-frontend

We also share the api at

• https://ecommute.micorix.dev/api/predict_pm10/day?l_spared=

  • where <n> is number of trips not taken
  • output: difference between predicted pm10 level and level of pm10 due to the number of trips not taken at given day.

• https://ecommute.micorix.dev/api/predict_pm10/day?l_spared=&hour=

  • where <n> is number of trips not taken. <h> is hour - int from range [0, 23]
  • output: difference between predicted pm10 level and level of pm10 due to the number of trips not taken at given hour.