def train(Xtrain, Ytrain, fmodel, isfile=True):
isfile = bool(int(isfile))
if isfile:
Xtrain = pd.read_csv(Xtrain)
Ytrain = pd.read_csv(Ytrain)
Xtrain = rain_feature.filter_features(Xtrain)
Xtrain = Xtrain.set_index('Id')
Xtrain[np.isnan(Xtrain)] = 9999
Xtrain[np.isinf(Xtrain)] = 9999
Ytrain = Ytrain.set_index('Id')
Ytrain.fillna(9999, inplace=True)
# Needed for SciKit Gradient Boosting Regressor
print 'Read the input: %d' % len(Xtrain)
learning_rate = 0.1
ntrees = 1000
num_features = 5
max_depth = 8
alpha = 0.9
v = 1
gbr = GBR_train(Xtrain, Ytrain, learning_rate, ntrees, num_features,
max_depth, alpha)
gbr_save_model(gbr, fmodel)
# Needed for XGBoost
'''
def cross_validate(Xtrain, Ytrain, op_file='run.cv', isfile=True):
isfile = bool(int(isfile))
if isfile:
Xtrain = pd.read_csv(Xtrain)
Ytrain = pd.read_csv(Ytrain)
Xtrain = rain_feature.filter_features(Xtrain)
Xtrain = Xtrain.set_index('Id')
Xtrain[np.isnan(Xtrain)] = 9999
Xtrain[np.isinf(Xtrain)] = 9999
Ytrain = Ytrain.set_index('Id')
Ytrain.fillna(9999, inplace=True)
# Run SciKit Cross Validation for Gradient Boosting Regressor
GBR_cv(Xtrain, Ytrain, op_file)
# Needed for XGBoost
'''
def cross_validate(Xtrain,Ytrain,op_file='run.cv',isfile=True):
isfile = bool(int(isfile))
if isfile:
Xtrain = pd.read_csv(Xtrain);
Ytrain = pd.read_csv(Ytrain);
Xtrain = rain_feature.filter_features(Xtrain);
Xtrain = Xtrain.set_index('Id');
Xtrain[np.isnan(Xtrain)] = 9999;
Xtrain[np.isinf(Xtrain)] = 9999;
Ytrain = Ytrain.set_index('Id');
Ytrain.fillna(9999,inplace=True);
# Run SciKit Cross Validation for Gradient Boosting Regressor
GBR_cv(Xtrain,Ytrain,op_file);
# Needed for XGBoost
'''
def train(Xtrain,Ytrain,fmodel,isfile=True):
isfile = bool(int(isfile));
if isfile:
Xtrain = pd.read_csv(Xtrain);
Ytrain = pd.read_csv(Ytrain);
Xtrain = rain_feature.filter_features(Xtrain);
Xtrain = Xtrain.set_index('Id');
Xtrain[np.isnan(Xtrain)] = 9999;
Xtrain[np.isinf(Xtrain)] = 9999;
Ytrain = Ytrain.set_index('Id');
Ytrain.fillna(9999,inplace=True);
# Needed for SciKit Gradient Boosting Regressor
print 'Read the input: %d'%len(Xtrain);
learning_rate = 0.1; ntrees = 1000; num_features = 5; max_depth = 8; alpha = 0.9; v = 1;
gbr = GBR_train(Xtrain, Ytrain, learning_rate, ntrees, num_features, max_depth, alpha);
gbr_save_model(gbr,fmodel);
# Needed for XGBoost
'''
def predict(fmodel, testX, op_file='data/output.csv', isfile=True):
if isfile:
testX = pd.read_csv(testX)
testX = rain_feature.filter_features(testX)
testX = testX.set_index('Id')
testX[np.isnan(testX)] = 9999
testX[np.isinf(testX)] = 9999
print "Predicting"
'''
bst = xgboost_load_model(fmodel);
xg_val = xgb.DMatrix(testX.as_matrix(),missing=np.nan);
pred = bst.predict(xg_val);
'''
gbr = gbr_load_model(fmodel)
pred = GBR_predict(gbr, testX)
pred = pd.DataFrame(pred)
pred.index = range(1,
len(pred) + 1)
pred.columns = ['Expected']
pred.index.name = 'Id'
pred.to_csv(op_file)
def predict(fmodel,testX,op_file='data/output.csv',isfile=True):
if isfile:
testX = pd.read_csv(testX);
testX = rain_feature.filter_features(testX);
testX = testX.set_index('Id');
testX[np.isnan(testX)] = 9999;
testX[np.isinf(testX)] = 9999;
print "Predicting";
'''
bst = xgboost_load_model(fmodel);
xg_val = xgb.DMatrix(testX.as_matrix(),missing=np.nan);
pred = bst.predict(xg_val);
'''
gbr = gbr_load_model(fmodel);
pred = GBR_predict(gbr,testX);
pred = pd.DataFrame(pred);
pred.index = range(1,len(pred)+1);
pred.columns = ['Expected'];
pred.index.name = 'Id';
pred.to_csv(op_file);
