Analyzing/Predicting CO2 emissions by vehicles with different features
-
Scrapped model-specific vehicle data for over 10 years from the Govt. of Canada’s open data website and performed exploratory data analysis (EDA) and feature engineering.
-
Curated a multivariate regression model using group LASSO and heavyLm to predict CO2 emissions. The model was able to forecast CO2 emissions with a mean absolute error (MAE) of 19.29 grams/kilometer and could be useful in implementing federal regulations to combat climate change.