W9D2

Random Forests etc.

Decision trees and Random forests are very powerful classication techniques and today we will use them on the Iris dataset. The tasks are as follows:

1. After conducting data exploration, write a formal hypothesis on which is the most significant variable.
2. Classifiy using the decisiontreeclassifier, randomforestclassifier and logisticregression.
3. Tune the hyperparameters in the decisiontreeclassifier to get the best possible results.
4. For the decision tree classifier, print out 2 graphs using graphviz (Mac: brew install graphviz, Windows: conda install python-graphviz) for the classifier without tuning and one with tuning.
5. For each classifer, plot the decision boundry (example: http://scikit-learn.org/stable/auto_examples/ensemble/plot_forest_iris.html#sphx-glr-auto-examples-ensemble-plot-forest-iris-py)
6. Write down the interpretation of the results