File = Bankdata.csv

URL : https://docs.google.com/spreadsheets/d/1Qn4pdGrhzLjoXy1F871BalEz57rqdsB8OvLd4AbhCbA/edit#gid=1783727779

Data Preprocessing

Q. Explore the dataset to identify the features and the class attribute. In general, scikit-learn doesn't deal with categorical data well. Some classifiers need normalized data. Consider if there are any missing values,outliers, and attributes that have no predict power.

Q. Convert pandas DataFrames into numpy arrays that can be used by scikit-learn. Show your data after being preprocessed. If none of the techniques described below is able to achieve close to 90% accuracy, examine your data again to see if you can preprocess the data in a different way.

Apply the following techniques to your preprocessed data set, and see which one yields the highest accuracy as measured with 10-fold cross validation.

Decision tree

Q. Create a single train/test split of your data. Set aside 75% for training, and 25% for testing. Use tree.DecisionTreeClassifier to create a model and fit it to your training data. Measure the accuracy of the resulting decision tree model using your test data. (Hint: you don't have to visualize the tree and you can use score method to get the accuracy.)

Q. Instead of a single train/test split, use 10-fold cross validation to get a measure of your model's accuracy. (Hint: use model_selection.cross_val_score and use mean method to find the average)

Random forest

Q. Use ensemble.RandomForestClassifier with n_estimators=10 and use 10-fold cross validation to get a measure of the accuracy. Does it perform better than decision tree?