1) Randomly divide the data into 80% for training and 20% for testing. Apply the following:
a) Handle the missing values in both train and test set.
b) Encode categorical variables using appropriate encoding method (in-built function allowed).
c) After completing step (a) and (b), compute 5-fold cross validation on the training set
(normalisation of data is allowed, if required). Print the final test accuracy.
2) Apply PCA (select number of components by preserving 95% of total variance) on the processed data from step (1).
a) Plot the graph for PCA (in-built function allowed for PCA and visualisation).
b) Use the features extracted from PCA to train your model. Compute 5-fold cross validation on the training set (normalisation of data is allowed, if required). Print the final test accuracy.
3) Using the processed data from step (1), apply the following:
a) A feature value is considered as an outlier if its value is greater than mean + 3 x standard deviation. A sample having maximum such outlier features must be dropped.
b) Using the sequential backward selection method, remove features.
c) Print the final set of features formed.
d) Compute 5-fold cross validation on the training set (normalisation of data is allowed if required). Print the final test accuracy. Report and results.
