Skip to content

A supervised classification machine learning approach to forecasting the road as safe (label 1) or dangerous (label 0) for driving in the arctic regions. If the friction is 0 <= x < 0.5 then we labeled it as 0, either 1 in the range 0.5 to 1.

License

Notifications You must be signed in to change notification settings

Abrar2652/Road-Friction-Forecasting

Repository files navigation

Road-Friction-Forecasting

The Colab Notebook link of this project: https://colab.research.google.com/drive/1W15eOQbeHp9wbJWRaE0f7ZfYv_jAj14O?usp=sharing

View this project on nbviewer (if you face any problem opening Jupyter Notebook): https://nbviewer.jupyter.org/github/Abrar2652/Road-Friction-Forecasting/blob/master/Road_Friction_Forecasting.ipynb

Dataset Collection

The dataset has been collected from the Smart Road - Winter Road Maintenance Challenge 2021 organized by UiT The Arctic University of Norway on Devpost. Hackathon link: https://dit4bears.devpost.com

Dataset download link: https://uitno.app.box.com/s/bch09z27weq0wpcv8dbbc18sxz6cycjt

After downloading the smart_road_measurements.csv file from the competition page, we had added extra columns collecting data from the external resources authorized the organizers. The links of the external datasets are:

[1] Weather data https://pypi.org/project/wwo-hist/

[2] UV Index data https://pyowm.readthedocs.io/en/latest/v3/uv-api-usage-examples.html

Technologies Used

  1. Python
  2. Jupyter Notebook
  3. Microsoft Excel
  4. Pandas
  5. XGBooster regressor
  6. Matplotlib
  7. Seaborn heatmap
  8. Optuna

About

A supervised classification machine learning approach to forecasting the road as safe (label 1) or dangerous (label 0) for driving in the arctic regions. If the friction is 0 <= x < 0.5 then we labeled it as 0, either 1 in the range 0.5 to 1.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published