Auto MPG
A Streamlit app on the Auto MPG dataset (398 vehicles, 1970 to 1982), built as an EMSI assignment. One tab explores the data without any machine learning: filters, averages, the weight-to-consumption trend and comparisons by origin. The other predicts fuel consumption in L/100 km with a random forest.
14 / 14 · Open source · data analysis and prediction · EMSI · 2026
What I did
- Descriptive analysis tab with filters by origin, cylinders, year and minimum mpg
- RandomForestRegressor (300 trees) predicting L/100 km from seven vehicle features