ML algorithms such as boosting, support vector machines or deep learning models could provide high flexibility and accuracy in predicting future outcomes when trained on a complex dataset. However, as the flexibility of the model increases, its interpretability is often lost which prevents us to derive useful relations between the predictors and our predictions. In some practical applications, we may require this information between the features and the outcomes, i.e we are interested in inference: the relationship between the features and the target values. In this context, one might prefer to work with simple and interpretable models in the context of regression and classification type tasks in order to better capture such relationships. To make myself more familiar with the Python Data Science stack, I have been deriving solutions (see the github repo) for the exercises of the Introduction to Statistical Learning book. ISLP is perfect for this purpose as it provides an intuitive approach while keeping a good dosage of theoretical/mathematical background for the materials covered. The current progress of the project covers many fundamental concepts in data science/ML within the context of interpretable models (First 4 Chapters of ISLP) including various diagnostic tools to analyze the goodness of fit of ML models, their tuning for bias/variance trade-off and Bayesian statistics and probability in the context of classification using the standard Python data science libraries: numpy, pandas, statsmodels, scikit-learn, matplotlib and seaborn.