adding donow

[mercye]

this donow was a good review of a lot of things -- like connecting to databases, splicing arrays, making decision tree classifiers, handling plots in matplotlib.

the most difficult thing for me is still connecting to databases, seamlessly. it requires some tinkering around before i know how to do it. i realized i am confused about whether 10 fold validation means 10 splits on the data that are then split into training and testing data sets OR five splits on the whole data set that are then split into training and testing.

i learned about .ravel() and the necessity of having your target array having only rows (x rows,) rather than being shaped like (x rows, 1) if you want to use cross_val_scores.