Download Balance-Scale and Flags datasets from UCI Irvine
1. Summarize the datasets in your words. What features (attributes) are used? [5 Points]
2. Using python and KNN, NaiveBayes, and C4.5 classi ers. Compare and analyse the performance using F-measure and Accuracy. Plot performance curves and discuss. Use 10 Fold Cross Validation and random train/test split(70%, 30%) [20 Points]
3. Repeat using SVM classi er. Use RBF Kernel with Chi-squared distance metric: Chi-squared distance metric: k(xi; xj) = e A1 dist 2 (xi;xj) where A is a scalar which normalises the distances. A is set to the average 2 distance between all elements of the kernel matrix. For simplicity, use only random train/test split(70%, 30%). [25 Points]
1
