Python, Machine Learning, Decision trees, Data. Задание 4 курса для университета.

Нужен только первый проект из документа. Вся работа на Python. Закончить надо 17 ноября в 10:30 вечера по МСК.
(1) Implement four (4) decision tree (DT) classifiers that use the information gain, gain ratio, variance and Gini index as split criteria, respectively. 
Note that you can use either binary-split or multiple-split; you can also use any reasonable pre-processing method and any reasonable early stopping criteria (e.g. pre-pruned parameters) but need to explain your reasons. (you may use ML libraries)
(2) Implement an ensembled DT classifier M* whose predictions are based on a voting function between the three classifiers, (you need to implement this task in Python, you cannot use a library). 
(3) Implement a 10-fold cross-validation to evaluate M* and Information Theory based decision tree, and perform Student’s t-test to determine the statistical significance of the error rate difference between the two classifiers. Note that you should set a suitable significance level. (you need to implement the ensemble model by Python code and you cannot use a library for this)
(4) Perform a full evaluation using the four methods and M* and various pruning parameters. Report the TP rates and FP rates for both the unpruned and the pruned models. For pruning use 1/3 dataset for training, 1/3 for pruning and 1/3 for evaluation. The three subsets must be generated by using stratified sampling. You need to implement the stratified method yourself.