Update Day 13 SVM.md

[bibin-johny]

Is that right to use sc.transform in Feature Scaling? I think it is sc.fit_transform()

[addy086]

Thank

[PatrickRuan]

I added and modified Day13 code a little below,
1.) added a "print out" for calculating error number at test data.
print('#error@test: ', cm[0,1]+cm[0,1])

2.) used an rbm kernel for SVM,
classifier = SVC(kernel ='rbf', random_state = 0) it helps day 16.

3.) used an poly kernel for SVM, the result won't better than linear: degree=2,..,10
classifier = SVC(kernel = 'poly',degree=3, random_state = 0)
but, try degree = 2, might prive a good practice for beginners: what spaces make data projection like this...