def main():
combined = utils.load_data()
combined = utils.drop_stupid(combined)
combined = utils.cat_transform(combined, type='onehot')
combined.replace(-1, combined.median(axis=0), inplace=True)
combined = utils.data_transform(combined, 'log')
train, test = utils.recover_train_test_na(combined, fillna=False)
# Prepare X and y following with parsing to numpy datatype
X_train = train.drop('target', axis=1).values
y_train = train.target.values
X_test = test.values
# Train Data
params = {
'class_weight': 'balanced',
'verbose': 1
}
model = LogisticRegression(**params)
# Start trainning
print('Ready to train with:')
print('Model name ', MODEL_NAME)
print('Model parameters ', model)
print('X_train shape is', X_train.shape)
print('y_train shape is', y_train.shape)
print('X_test shape is', X_test.shape)
layer1.make_oof(model, X_train, y_train, X_test, MODEL_NAME)
def train(self):
combined = utils.load_data()
if self.drop_stupid:
combined = utils.drop_stupid(combined)
# if self.bojan_features:
# combined = utils.bojan_engineer(combined)
if self.engineer_stats:
combined = utils.engineer_stats(combined)
if self.recon_category:
combined = utils.recon_category(combined)
if self.cat_transform:
combined = utils.cat_transform(combined, self.cat_transform)
if self.data_transform:
combined = utils.data_transform(combined, self.data_transform)
if self.feature_interactions:
combined = utils.feature_interactions(combined)
if self.kinetic_feature:
combined = pd.concat([
combined,
pd.read_csv('data/kinetic_combined.csv', index_col='id')
],
axis=1)
if self.my_features:
calc = utils.load_data()
calc = calc[calc.columns[calc.columns.str.contains('calc')]]
calc = pd.get_dummies(calc, columns=calc.columns)
calc = calc[[
'ps_calc_02_0.0', 'ps_calc_02_0.1', 'ps_calc_05_3',
'ps_calc_06_7', 'ps_calc_06_10', 'ps_calc_07_5',
'ps_calc_08_8', 'ps_calc_08_10', 'ps_calc_10_8',
'ps_calc_11_5', 'ps_calc_11_7', 'ps_calc_11_8'
]]
combined = pd.concat([combined, calc], axis=1)
train, test = utils.recover_train_test_na(combined, fillna=self.fillna)
X_train = train.drop('target', axis=1)
y_train = train.target
X_test = test
# Start trainning
print('Ready to train with:')
print('Model name ', self.MODEL_NAME)
print('Model parameters ', self.model)
print('X_train shape is', X_train.shape)
print('y_train shape is', y_train.shape)
print('X_test shape is', X_test.shape)
self.make_oof(self.model, self.params, X_train, y_train, X_test,
self.MODEL_NAME)
def train(self):
log_path = os.path.join(LOG_PATH, self.MODEL_NAME + '_log.txt')
orig_stdout = sys.stdout
f = open(log_path, 'w')
sys.stdout = f
combined = utils.load_data()
if self.kinetic_transform:
combined = utils.kinetic_transform(combined)
if self.drop_stupid:
combined = utils.drop_stupid(combined, type=self.drop_stupid)
if self.engineer_stats:
combined = utils.engineer_stats(combined)
if self.recon_category:
combined = utils.recon_category(combined)
if self.cat_transform:
if self.cat_transform != 'smooth':
combined = utils.cat_transform(combined,
type=self.cat_transform)
if self.data_transform:
combined = utils.data_transform(combined, type=self.data_transform)
if self.feature_interactions:
combined = utils.feature_interactions(combined)
train, test = utils.recover_train_test_na(
combined, remove_outliers=self.remove_outliers)
X_train = train.drop('target', axis=1)
y_train = train.target
X_test = test
# Start trainning
print('Ready to train with:')
print('Model name ', self.MODEL_NAME)
print('Model parameters ', self.model.get_params())
print('X_train shape is', X_train.shape)
print('y_train shape is', y_train.shape)
print('X_test shape is', X_test.shape)
self.make_oof(self.model, X_train, y_train, X_test, self.MODEL_NAME)
sys.stdout = orig_stdout
f.close()
from keras.models import load_model, Sequential
from keras.models import Sequential, Model
from keras.layers import Input, Dense, Dropout, Activation, Reshape, Concatenate, Merge
from keras.layers.normalization import BatchNormalization
from keras.callbacks import EarlyStopping, ModelCheckpoint, Callback, CSVLogger
from keras.wrappers.scikit_learn import KerasClassifier
from keras.optimizers import SGD
from keras.layers.embeddings import Embedding
np.random.seed(88) # for reproducibility
MODEL_NAME = 'keras_smooth_itt'
SEED = 88
combined = utils.load_data()
# combined = utils.bojan_engineer(combined)
combined = utils.drop_stupid(combined)
combined = utils.engineer_stats(combined)
combined = utils.recon_category(combined)
combined = utils.minmaxpandas(combined)
combined = combined.replace(np.NaN, -1)
# combined = utils.cat_transform(combined, 'onehot')
# combined = utils.data_transform(combined, self.data_transform)
# combined = utils.feature_interactions(combined)
train, test = utils.recover_train_test_na(combined, fillna=False)
X_train = train.drop('target', axis=1)
y_train = train.target
# X_test = test
print('\n')
from keras.layers.normalization import BatchNormalization
from keras.callbacks import EarlyStopping, ModelCheckpoint, Callback, CSVLogger
from keras.wrappers.scikit_learn import KerasClassifier
from keras.optimizers import SGD
from keras.layers.embeddings import Embedding
from tensorflow import set_random_seed
set_random_seed(MODEL_SEED)
np.random.seed(MODEL_SEED) # for reproducibility
MODEL_NAME = 'keras_joe_itt'
SEED = MODEL_SEED
combined = utils.load_data()
# combined = utils.bojan_engineer(combined)
combined = utils.drop_stupid(combined, 'default')
combined = utils.engineer_stats(combined)
# combined = utils.cat_transform(combined, 'onehot')
# combined = utils.data_transform(combined, self.data_transform)
# combined = utils.feature_interactions(combined)
train, test = utils.recover_train_test_na(combined, fillna=False)
# Fillna for minmax scaler
train = train.replace(np.NaN, -1)
test = test.replace(np.NaN, -1)
X_train = train.drop('target', axis=1)
y_train = train.target
X_test = test
col_vals_dict = {