Advanced programming

This section deepens students' Python programming skills beyond the foundations covered in Modules 1 and 2. The module opens with a consolidation of string formatting techniques, including f-strings, before moving into one of the most important paradigms in modern software development: Object-Oriented Programming (OOP). Students learn how to define classes, create objects, and write methods — concepts that underpin much of the Python code used in data science and machine learning libraries. The module then expands the Python toolkit with additional data structures (tuples and sets) and introduces file handling: how to read input from files and write output to files, including an understanding of file objects as consumable resources. Throughout, students apply and consolidate these skills through hands-on exercises and by annotating their own data science project notebooks.

From Preprocessing to Prediction: Building a Machine Learning Pipeline

This section guides students through the complete sequence of steps that transform raw data into a working classification model. Building on the preprocessing introduced in Module 2, students consolidate their understanding of handling missing values and duplicates, then move on to categorical data encoding — exploring the different strategies available and the reasoning behind each choice. A dedicated focus is placed on class imbalance: how to detect it, interpret it, and decide among the available strategies for addressing it. Students then learn the principles and good practices of splitting data into training and test sets, including stratification and random seed reproducibility, before confronting a critical and often underestimated issue: data leakage and the importance of fitting transformations exclusively on training data. The module culminates in the creation, fitting, and evaluation of a first classification model using K-Nearest Neighbors (KNN), introducing the conceptual framework of true/false positives and negatives as the foundation for model evaluation metrics. Students apply every step to their own dataset, progressively building an annotated, reproducible machine learning workflow.