get_data.py from_2010(path, tag, train_hour, max_train_hour, pred_hour) to get raw feature e.g. temperature tag is the column in number

get_time(path, max_hour, pred_hour)
	to return dade and time been normalized by 0~1
	date is divided by 366
	time is devided by 24

insert.py to fill in null features

train.py to train model the feature may be raw or been normalized

train_cross.py to do cross validation and testing for each year

train_PCA.py the feature been PCAed

tuning.py to tune parameters with raw feature or normalized

tuning_PCA.py to tune parameters with PCAed feature