def predict_model(train_x, train_y, test_x, test_y=None, class_weights=None, random_state=0):
# Learn the ProbMC model
params = {'n_jobs':-1, 'random_state':random_state,
'n_estimators':1000, 'max_features':0.75, 'min_samples_leaf':1}
model = fp.PMC_MultiTaskExtraTreesRegressor(train_y.shape[1], bags=1, params=params)
sample_weight = None
model.fit(train_x, train_y, sample_weight)
# Predict on the test instances
test_predicted = model.predict_proba(test_x)
# Learn the ProbMC model
params = {'n_jobs':12, 'random_state':random_state,
'n_estimators':1000, 'max_features':0.45, 'min_samples_leaf':1}
model = fp.PMC_MultiTaskExtraTreesRegressor(train_y.shape[1], bags=1, params=params)
sample_weight = None
model.fit(train_x, train_y, sample_weight)
# Predict on the test instances
test_predicted = test_predicted + model.predict_proba(test_x)
return(test_predicted / 2)
def predict_model(train_x,
train_y,
test_x,
test_y=None,
class_weights=None,
random_state=0):
# Learn the ProbMC model
params = {
'n_jobs': -1,
'random_state': random_state,
'bootstrap': False,
'n_estimators': 250,
'max_features': 0.25,
'min_samples_leaf': 3
}
model = fp.PMC_MultiTaskExtraTreesRegressor(train_y.shape[1],
bags=1,
params=params)
sample_weight = np.dot(np.power(train_y + 1, -1),
np.power(class_weights, -1))
model.fit(train_x, train_y, sample_weight)
# Predict on the test instances
test_predicted = model.predict_proba(test_x)
return (test_predicted)