Project information

  • Problem: Air Quality Forecasting
  • ML Areas: Multi-Variate Time Series Forecasting
  • Learning technique: Supervised Learning
  • Tools: Python, Jupyter Notebook
  • Project date: April 2020
  • Project URL: AirQo-Challenge project

Air Quality Forecasting Project

Licence

Python Jupyter

Libraries:

catboost xgboost Scikit_learn Pandas

Contents

Description

Air quality prediction for Uganda at exactly 24 hours after a 5-day series of hourly weather data readings which include temperature, rainfall, wind, and humidity. This project is based on Zindi competition. Here you can read the full description of the challenge

Dataset

Dataset details available at zindi challenge data

Task

Multi-Variate Time-Series Forecasting

Solutions

  • Classic ML Algorithms

    • Linear Regression
    • Ridge Regression
    • Lasso Regression

  • Ensemble Algorithms

    • Random Forest
    • Catboost
    • XGBoost

Evaluation

  • RMSE

Result

XGBoost: RMSE = 37.95

competition_certification

Author

Daniele Moltisanti

Linkedin