1. UCI ML Repository contains many datasets for classification. You need to find 5 datasets with at least 10 attributes
https://archive.ics.uci.edu/ml/datasets.php
Complete the following tables and calculate accuracy using (1) Use 10 x 10 Fold CV (ii) 70% Holdout approach repeated 100 times
Show all the standard deviations in the table. Briefly discuss advantages / disadvantages of hold out and cross validation approach. Analysis the result. Which approach is good and why? Why some approaches unable to perform well in some data sets.
Dataset1 Dataset2
Dataset3
Dataset4
Dataset5
DT using gini
(without pruning)
DT using gini
(with pruning)
DT using entropy
(without pruning)
DT using entropy
(with pruning)
Hint: Check ccp_alpha parameter for pruning. Use ccp_alpha = 0.015 for pruning
