The challenge was to optimize the evaluation metrics, by testing out different models on the truck fleet dataset. L1 regularization was performed on the dataset to obtain the important features, and was then followed by logistic regression model to train the dataset.
Before settling onto the logistic model, several other models were tested, but when performing feature selection using L1 regularization, the best fit was to run the logistic model.
Led to reduction of misclassification rate by 17.18%, increased the AUC by 38.18% and decreased the RMS by 9.96%.