In part II of this assigment use a multivariate feature selection
method to rank and select features. Specifically, run the linear
SVM first to determine a w vector. Then rank the features by the
absolute value of entries in w (in other words |wi|). 

Instead of ranking all features with the SVM first reduce the features
to top 50 and then rank the 50 with an SVM.

Don't forget to standardize the columns of the train dataset
before doing multivariate feature ranking. Otherwise the w vector
will have large entries if a column has many 2's and 1's compared
to a column just a few 1's. We standardize a column by dividing
it by the Euclidean length.

Use the same training and test dataset as part I. 

Submit your assignment by copying it into the directory
/afs/cad/courses/bnfo/f18/bnfo/615/001/<ucid>
For example if your ucid is abc12 then copy your solution into 
/afs/cad/courses/bnfo/f18/bnfo/615/001/abc12

Your completed assignment is due on December 15th 2018.