Python Please...

4. Decision Tree

4.1 Initialize a decision tree model using sklearns DecisionTreeClassifier. Use the unbalanced training set. Set a consistent value for random_state parameter so that your result is reproducible.

4.2 Use grid search to find out the best combination of values for the parameters: criterion, max_depth, min_samples_split, max_features. Then print the best performing parameters.

4.3 Add the best performing parameter set to the already-initialized Decision Tree model. Then fit it on the train dataset.

4.4 Import the accuracy_score, precision_score, recall_score, confusion_matrix, f1_score, roc_auc_score from scikitlearn's metrics package. Evaluate your Decision Tree on the Test dataset and print all the metrics.

4.5 Plot the tree using scikitlearn's tree package. You may need to define a large figure size using matplotlib to have an intelligible figure.

4.6 Initialize a new Decision Tree model, then use the best set of parameters from Step 4.3 to train it on the balanced train set that you prepared in Step 3.5.1.

4.7 Print the evaluation scores (accuracy_score, precision_score, recall_score, confusion_matrix, f1_score, roc_auc_score) from the training on balanced dataset.

4.8 Discuss any difference between evaluation results from the unbalanced train set and balanced train set.