Overview

Schedule

Title	Description
Lecture 1: Course introduction	What is machine learning, types of machine learning, learning to navigate through the course materials, getting familiar with the course policies
Lecture 2: Terminology, baselines, decision Trees	Supervised machine learning terminology: Features, target, examples, training, parameters and hyperparameters, Decision boundary, classification vs. regression, inference vs. prediction, accuracy vs. error, baselines, intuition of decision trees
Lecture 3: ML fundamentals	Generalization, data splitting, cross-validation, overfitting, underfitting, the fundamental tradeoff, the golden rule
Lecture 4: \(k\)-nearest neighbours and SVM RBFs	Introduction to KNNs, hyperparameter `n_neighbours` or \(k\), `C` and `gamma` hyperparameters of SVM RBF, decision boundaries with different values of hyperparameters.
Lecture 5: Preprocessing and sklearn pipelines	Preprocessing motivation, Common transformations in `sklearn`, `sklearn` transformers vs. Estimators, The golden rule in the feature transformations, `sklearn` pipelines
Lecture 6: `sklearn` column transformer and text fearutres	Column transformer, arguments of OHE, encoding text features, incorporating text features in an ML pipeline
Lecture 7: Linear models	Intuition behind linear models, linear regression and logistic regression, `scikit-learn`’s Ridge model, prediction probabilities, interpret model predictions using coefficients learned by a linear model, parametric vs. non-parametric models
Lecture 8: Hyperparameter Optimization	motivation for hyperparameter optimization, hyperparameter optimization using sklearn’s GridSearchCV and RandomizedSearchCV, optimization bias
Lecture 9: Classification metrics	confusion metrics, precision, recall, f1-score, PR curves, AP score, ROC curve, ROC AUC, class imbalance
Lecture 10: Regression metrics	Classification metrics recap, `ridge` and `RidgeCV`, `alpha` hyperparameter of `ridge`, MSE, RMSE, MAPE, log transformations on the target
Lecture 11: Ensembles
Lecture 12: Feature importances
Lecture 13: Feature engineering and selection	Motivation for feature engineering, preliminary feature engineering on text data, general concept of feature selection, model-based feature selection vs. RFE
Lecture 14: K Means	Unsupervised paradigm, motivation and potential applications of clustering, K-Means algorithm, pros and cons of K-Means, the Elbow plot and Silhouette plots for a given dataset, importance of input data representation in clustering.
Lecture 15: DBSCAN and hierarchical	Limitations of K-Means, DBSCAN, ierarchical clustering, dendrograms, omparing and contrasting different clustering methods
Lecture 16: Recommender systems