NES ML-I Course Home Assignments

Practical home assignments of the base course "Machine Learning, part 1" (New Economic School, 2023). Topics per assignment:

1. Intro to pandas and matplotlib (used plotly instead)
2. EDA and linear regression models Ridge and Lasso on New York City Taxi Trip Duration problem data, alpha grid search
3. Self-built gradient descent and variations: (batch) SGD, Momentum, AdaGrad, RMSProp. Tested on data from the second assignment
4. Binary classification problem, usage of LogisticRegression and SVC. Calibration curves, interpretation of scores. Different ways of encoding categorical variables: OrdinalEncoder, OneHotEncoder, TargetEncoder. Different ways of feature selection (after OHE): embedded methods, filet methods, wrapper methods.
5. Decision trees (classification), application of DecisionTreeClassifier. Self-constructed tree classifier, Gini index. Effects of max depth, min samples in leaf, and min samples in split.
6. Self-built gradient boosting. Hyperparameter search, usage of Optuna package. Usage of CatBoostClassifier, importance of features.