The course navigates through the basic concepts and principles behind the main data science models and techniques. Descriptive techniques such as clustering and frequent pattern mining are explained in more details while, in case of predictive techniques, the focus is put mainly on the concepts of a model, its parameters and hyper-parameters as well as the quality and validation of models including overfitting-underfitting and the bias.-variance trade-offs. Data quality and pre-processing issues related to various data types and modeling problems are also tackled. Finally, basic recommendation techniques and the CRISP-DM methodology are contained in the course as well.

• Clustering: k-means, agglomerative, DBSCAN, cluster validation;
• Frequent Pattern Mining: itemsets, association rules, quality measures;
• Linear Classification and Regression: model, parameters and hyper-parameters, validation, overfitting-underfitting and the bias-variance trade-off;
• Introduction to traditional prediction techniques (as black-box functions);
• data quality and pre-processing: noise, missing values, data transformation, normalization;
• the CRISP-DM process;
• recommendation techniques;