dc.green_groupintodailydf(paths, 'Green' + str(year) + 'DailyDrivers.csv', printcsv)
elif color == 'yellow':
# Group
dc.yellow_groupintodailydf(paths, 'Yellow' + str(year) + 'DailyDrivers.csv', printcsv)
if __name__ == "__main__":
#cleanandgroup(color='yellow', year=2015, nummonths=3, clean=False, printcsv=True)
# Predict
df = pd.read_csv('Yellow2014DailyDrivers.csv', usecols=[0,1,2,3,4,5])
p = TuneModel(df,'Day_taxi_drivers', testsize=.25, modeltype='regress', impute=False, scores=['r2'])
clf = p.randomforestreport(folds=5, cores=1, plotit=False, saveit=False)
# Predicting into 2015 (using all of 2014 as the training set)
# clf = RandomForestRegressor(n_estimators=250)
# #todo: make this into function, where number of days in X_test is a parameter
# y_train = df['Day_taxi_drivers']
# X_train = df.drop('Day_taxi_drivers', axis=1)
#
# # Create X_test set going forward three months (and add holidays in)
# X_test = modelutilities.create_future_X_test('01-01-15','03-31-2015',freq='D')
# holidaylist=['2015-01-01','2015-02-13','2015-02-14','2015-03-16','2015-03-17']
# X_test['Holiday'] = 0
# for i in holidaylist: # loc should be able to obviate forloop
# X_test.loc[i,'Holiday'] = 1
# o.notokenlogitreport(folds=2, cores=7)
#
# o.tokenlogitreport(folds=2, cores=7)
#
# o.notokensvcreport(folds=2, cores=-1)
#
# o.tokensvcreport(folds=2, cores=-1)
#### CSV Example of how to use TuneClassifier and printreport for hyperparameter optimization
df = pd.read_csv('SalesOrderHeaderNULL.csv')
#p = PredictListFactors(df, 'OnlineOrderFlag', 'SalesPersonID', 'ShipMethodID')
#p.predictfactors()
p = TuneModel(df,'OnlineOrderFlag', testsize=.5, type='regress')
#
p.logitreport(folds=2, cores=6)
#
# p.treesreport(folds=2, cores=6)
#
# p.extratreesreport(folds=2, cores=6)
#
# p.randomforestreport(folds=2, cores=6)
## CSV example of how to use OverallRank class and methods printit, plotit
## General model feature importance
#t = OverallRank(df,'OnlineOrderFlag')
#t.printit()