This is a prediction of the likelihood of a ticket being paid, using a real-life dataset. It was done for a Coursera project. The model building involved data exploration and cleaning (using pandas), feature engineering (selction, transformation), model training and validation. Two binary classifiers were developed - a linear classifier (logistic regresssion w/ L1 regularalizer) and a non-linear classifier (gradient-boosted decision trees). The hyperparameters were tuned using the validation set. The model's ROC-AUC was close to 80%, which is close to the highest that has been obtained on this data set.