Final

The theme of my bachelor thesis is interpretability of multi-Label classification problem. Multi label classification problem means that each sample may belong to more than one class. And interpreter can interpret the prediction result of classifiers. That is to say, it can show us the probability or index of a sample belongs to each class. The more interpretable a machine learning model is, the easier it is for people to understand why certain decisions or predictions are made. In my bachelor thesis I use Python to train and test the model with different classifiers, for example Binary Relevance Classifier, classifier chain and Multi-Label k-Nearest Neighbors. And I use different kinds of interpreters such as SHAP, LIME, Permutation Importance in Scikit-Learn and ELI5. The experience of writing this bachelor thesis gave me a deep understanding of machine learning especially classification problem and data analysis.