To predict wheather the payment type is cash or card from a vehicle which passed from a gateway.
To make the whole process clean and concise, I split up this assignment into 3 parts.
- part1: Feature engineering
- part2: Model Selection
- part3 Fit the chosen model and train/test dataset, make a prediction
The part1 explains what features I picked for training. It lists different combinations of features and finds out the set that will have the most accurate prediction. The part2 not only train 4 different and compare their performance, it also make a grid search to discover the best hyper parameters. Eventually, I chose decision tree because it's the most efficient algorithm. Finally, the part3 uses the outcomes from the previous 2 parts, applies the model on the test dataset and make a final prediction.
The prediction.csv is the result made from part3.