def train_model(self, ensembler=False):
params = {}
if ensembler:
params = get_model_params(ensembler, 'catboost')
else:
params = get_model_params()
self.model = CatBoostClassifier(iterations=params['iterations'],
learning_rate=params['learning_rate'],
depth=params['depth'],
l2_leaf_reg=params['l2_leaf_reg'],
od_type="Iter",
verbose=0)
if not (ensembler):
self.model.fit(self.X, self.y)
return self.model
def train_model(self, ensembler):
params = {}
if ensembler:
params = get_model_params(ensembler, 'rfa')
else:
params = get_model_params()
self.model = RandomForestClassifier(
n_estimators=params['n_estimators'],
criterion=params['criterion'],
max_features=params['max_features'],
max_depth=params['max_depth'],
n_jobs=-1,
random_state=42,
class_weight=params['class_weight'])
self.model.fit(self.X, self.y)
return self.model
def start_preprocessing(self):
print('\nCleaning up columns...')
model_params = get_model_params()
self.remake_date_cols()
self.remake_states_lga()
self.find_missing_val_replacements()
self.clean_gender_col(self.preproc_args['missing']['gender'])
self.clean_age_col(int(self.preproc_args['missing']['age']),
self.preproc_args['missing']['age_map'])
self.clean_color_col(self.preproc_args['missing']['color'])
self.clean_carmake_col(self.preproc_args['missing']['make'])
self.clean_carcat_col(self.preproc_args['missing']['category'])
#self.clean_car_product(self.preproc_args['missing']['product'])
#self.clean_state_col(self.preproc_args['missing']['state'])
#self.clean_LGA_col(self.preproc_args['missing']['lga'])
self.clean_date_col()
#self.remake_nopol_col()
print('\nPlotting distribution...')
self.plot_graph()
print('\nApplying label encoder...')
#self.find_feature_correlation(self.preproc_args['correlation_LB'],self.preproc_args['correlation_UB'])
self.select_best_features(model_params['feature_k'])
self.drop_skip_columns()
self.encode_labels()
return self.apply_oversampling()
def train_model(self, ensembler = False):
params = {}
if ensembler:
params = get_model_params(ensembler, 'lgbm')
else:
params = get_model_params()
#self.X = self.X.rename(columns = lambda x: re.sub('[^A-Za-z0-9_]+','',x))
self.model = LGBMClassifier(random_state = 42,
class_weight=params['class_weight'],
n_estimators = params['n_estimators'],
learning_rate = params['learning_rate'],
min_split_gain = params['min_split_loss'],
num_leaves = params['num_leaves'],
min_child_samples = params['min_child_samples'],
min_child_weight = params['min_child_weight'])
if not(ensembler):
self.model.fit(self.X, self.y)
return self.model
def train_model(self):
params = get_model_params()
if params['kernel'] != 'linear':
self.model = SVC(C=params['C'],
kernel=params['kernel'],
degree=params['degree'],
gamma=params['gamma'],
tol=params['tol'],
class_weight=params['class_weight'],
max_iter=-1,
random_state = 42)
else:
self.model = LinearSVC(C=params['C'],
loss=params['loss'],
penalty=params['penalty'],
class_weight=params['class_weight'],
random_state = 42,
max_iter = params['max_iter'],
dual = params['dual']
)
self.model = self.model.fit(self.X,self.y)
return self.model
def __init__(self, X, y):
self.X = X
self.y = y
self.model_params = get_model_params()
def read_args():
args = get_model_params()
val_args = get_validation_params()
main(args, val_args)
def setup_voting_classifier(self, models):
params = get_model_params()
model = VotingClassifier(estimators=models,
voting=params['voting_type'])
model.fit(self.X, self.y)
return model