def __init__(self, pipeline_):
train = pd.read_csv('data/train.csv', index_col=0)
test = pd.read_csv('data/test.csv', index_col=0)
add_features = AddFeatures(train, test)
add_features.add_bmi_sist_dist_map()
add_features.add_f_score()
add_features.add_ap_features()
add_features.del_features()
train = add_features.train
test = add_features.test
self.Y = train['cardio'].values
train.drop('cardio', axis=1, inplace=True)
self.X = train
self.test = test
self.pipeline = pipeline_
self.model = None
self.kf = None
self.results = pd.DataFrame()
from sklearn.ensemble import RandomForestClassifier
from hyperopt import fmin, hp, STATUS_OK, Trials, tpe
import warnings
from FEATURES import AddFeatures
warnings.filterwarnings('ignore')
pd.set_option('display.max_columns', 16)
pd.set_option('display.width', 1000)
np.random.seed(42)
train = pd.read_csv('data/train.csv', index_col=0)
test = pd.read_csv('data/test.csv', index_col=0)
add_features = AddFeatures(train, test)
add_features.add_bmi_sist_dist_map()
add_features.add_f_score()
train = add_features.train
test = add_features.test
Y = train['cardio'].values
train.drop('cardio', axis=1, inplace=True)
X = train.as_matrix()
def hyperopt_train_test(hpparams):
params_est = {
'n_estimators': int(hpparams['n_estimators']),
'criterion': 'gini',
'max_features': hpparams['max_features'],