Feature Scaling
This is usually the last part on data preprocessing as after this step, the data is not really legible and no human can spot any pattern in it. The basic goal of this step is to get all of data between -1 and 1.