Now that we have learned three (or four if you would say) commonly used and powerful regression algorithms and performance evaluation metrics, why don't we utilize all of these in solving our stock price prediction problem?
We have generated the features that we need earlier, and now we will continue with constructing the training set based on data from 1988 to 2014:
>>> import datetime>>> start_train = datetime.datetime(1988, 1, 1, 0, 0)>>> end_train = datetime.datetime(2014, 12, 31, 0, 0)>>> data_train = data.ix[start_train:end_train]
All fields in the dataframe data (defined in the code from the beginning section) except 'close' are feature columns, and 'close' is the target column:
