yTrainRegistered = yRegistered[train]
XTest = X[test]
yTestCasual = yCasual[test]
yTestRegistered = yRegistered[test]
clfCasual = ExtraTreesRegressor()
# selectorCasual = SelectKBest(score_func=f_regression,k=20)
# print [1+zero_based_index for zero_based_index in list(selectorCasual.get_support(indices=True))]
clfRegistered = ExtraTreesRegressor()
modelCasualETR = clfCasual.fit(XTrain, yTrainCasual)
modelRegisteredETR = clfRegistered.fit(XTrain, yTrainRegistered)
predictionsCasual = clfCasual.predict(XTest)
actualCasual = yCasual[test]
scoresCasualExtraTreesRegression.append(evaluation.rmsle(actualCasual, predictionsCasual))
predictionsRegistered = clfRegistered.predict(XTest)
actualRegistered = yRegistered[test]
scoresRegisteredExtraTreesRegression.append(evaluation.rmsle(actualRegistered, predictionsRegistered))
predictionsTotal = np.sum([predictionsCasual, predictionsRegistered], axis=0)
actualTotal = np.sum([yTestCasual, yTestRegistered], axis=0)
currentScore = evaluation.rmsle(actualTotal, predictionsTotal)
scoresTotalExtraTreesRegression.append(evaluation.rmsle(actualTotal, predictionsTotal))
if currentScore == min(scoresTotalExtraTreesRegression):
mdlExtraTreesRegressorCasual = modelCasualETR
mdlExtraTreesRegressorRegistered = modelRegisteredETR
clfAdaBoostCasual = RandomForestRegressor()
XTest = X[test]
yTestCasual = yCasual[test]
yTestRegistered = yRegistered[test]
clfCasual = ExtraTreesRegressor()
# selectorCasual = SelectKBest(score_func=f_regression,k=20)
# print [1+zero_based_index for zero_based_index in list(selectorCasual.get_support(indices=True))]
clfRegistered = ExtraTreesRegressor()
modelCasualETR = clfCasual.fit(XTrain, yTrainCasual)
modelRegisteredETR = clfRegistered.fit(XTrain, yTrainRegistered)
predictionsCasual = clfCasual.predict(XTest)
actualCasual = yCasual[test]
scoresCasualExtraTreesRegression.append(
evaluation.rmsle(actualCasual, predictionsCasual))
predictionsRegistered = clfRegistered.predict(XTest)
actualRegistered = yRegistered[test]
scoresRegisteredExtraTreesRegression.append(
evaluation.rmsle(actualRegistered, predictionsRegistered))
predictionsTotal = np.sum([predictionsCasual, predictionsRegistered],
axis=0)
actualTotal = np.sum([yTestCasual, yTestRegistered], axis=0)
currentScore = evaluation.rmsle(actualTotal, predictionsTotal)
scoresTotalExtraTreesRegression.append(
evaluation.rmsle(actualTotal, predictionsTotal))
if currentScore == min(scoresTotalExtraTreesRegression):
mdlExtraTreesRegressorCasual = modelCasualETR
kf = KFold(len(X), n_folds=10)
for train, test in kf:
XTrain = X[train]
yTrainCasual = yCasual[train]
yTrainRegistered = yRegistered[train]
XTest = X[test]
yTestCasual = yCasual[test]
yTestRegistered = yRegistered[test]
model1 = BernoulliRBM(n_components=80)
model1.fit(XTrain)
weights_1 = model1.components_
XTrain_stage1 = XTrain.dot(weights_1.T)
model2 = BernoulliRBM(n_components=40)
model2.fit(XTrain_stage1)
weights_2 = model2.components_
XTrain_stage2 = XTrain_stage1.dot(weights_2.T)
#perform linear regression
regrCasual = linear_model.LinearRegression()
regrCasual.fit(XTrain_stage2, yTrainCasual)
regrRegistered = linear_model.LinearRegression()
regrRegistered.fit(XTrain_stage2, yTrainRegistered)
predictionsCasual = regrCasual.predict(XTrain_stage2)
score = evaluation.rmsle(yTrainCasual, predictionsCasual)
print score
kf = KFold(len(X), n_folds=10)
for train, test in kf:
XTrain = X[train]
yTrainCasual = yCasual[train]
yTrainRegistered = yRegistered[train]
XTest = X[test]
yTestCasual = yCasual[test]
yTestRegistered = yRegistered[test]
model1 = BernoulliRBM(n_components=80)
model1.fit(XTrain)
weights_1 = model1.components_
XTrain_stage1 = XTrain.dot(weights_1.T)
model2 = BernoulliRBM(n_components=40)
model2.fit(XTrain_stage1)
weights_2 = model2.components_
XTrain_stage2 = XTrain_stage1.dot(weights_2.T)
#perform linear regression
regrCasual = linear_model.LinearRegression()
regrCasual.fit(XTrain_stage2, yTrainCasual)
regrRegistered = linear_model.LinearRegression()
regrRegistered.fit(XTrain_stage2, yTrainRegistered)
predictionsCasual = regrCasual.predict(XTrain_stage2)
score = evaluation.rmsle(yTrainCasual, predictionsCasual)
print score
