X = artgor_utils.reduce_mem_usage(X)
X_test = artgor_utils.reduce_mem_usage(X_test)
gc.collect()
if debug:
print('saving debug data...')
X.to_csv("../data/debug_data/debug_acsf.csv", index=False)
print("training models...")
result_dict_lgb = artgor_utils.train_model_regression(
X=X,
X_test=X_test,
y=y,
params=params,
folds=folds,
model_type='lgb',
eval_metric='group_mae',
plot_feature_importance=True,
verbose=100,
early_stopping_rounds=1000,
n_estimators=n_estimators,
res_filename=result_filename)
if not debug:
print("saving results...")
np.save(result_filename, result_dict_lgb)
print("making submission...")
sub['scalar_coupling_constant'] = result_dict_lgb['prediction']
sub.to_csv(sub_filename, index=False)
index_type_test = (test['type'] == type)
X_t = train.loc[index_type].copy()
X_test_t = test.loc[index_type_test].copy()
y_t = y[index_type]
print('X_test_t.shape ', X_test_t.shape)
print('X_t.shape ', X_t.shape)
print('Training...')
result_dict_lgb = artgor_utils.train_model_regression(
X=X_t,
y=y_t,
X_test=X_test_t,
params=params,
columns=best_features,
folds=folds,
n_folds=n_folds,
model_type='lgb',
eval_metric='group_mae',
verbose=50,
early_stopping_rounds=1000,
n_estimators=n_estimators,
res_filename=result_filename)
del X_t, X_test_t
gc.collect()
if not debug:
np.save(str(result_filename), result_dict_lgb)
X_short.loc[X_short['type'] == type, 'oof'] = result_dict_lgb['oof']
X_short_test.loc[X_short_test['type'] == type,
if oof_colomn == 'scalar_coupling_constant':
y = coupling_constant
else:
y = folder_data[oof_colomn]
folder_data[categorical_cols] = folder_data[
categorical_cols].astype('category')
print('test_shape', test[features].shape)
print('train shape', folder_data[features].shape)
result_dict = artgor_utils.train_model_regression(
X=folder_data,
X_test=test,
y=y,
params=params,
folds=folds,
columns=features,
model_type='lgb',
eval_metric='group_mae',
verbose=50,
early_stopping_rounds=20,
n_estimators=n_estimators,
res_filename=result_filename,
)
print("saving results...")
if result_filename:
np.save(str(result_filename), result_dict)
type_test = type_dist_test
gc.collect()
params = types_config[type]
result_filename = result_filename_prefix / f'distance_kernel_{type}.npy'
type_y = full_train[train_mask]['scalar_coupling_constant']
print('X_test_t.shape ', type_test.shape)
print('X_t.shape ', type_train.shape)
print('Training...')
result_dict_lgb = artgor_utils.train_model_regression(
X=type_train, y=type_y, X_test=type_test,
params=params,
columns=best_features,
folds=folds,
n_folds=n_folds,
model_type='lgb',
eval_metric='mae',
verbose=50,
early_stopping_rounds=1000,
res_filename=result_filename
)
del type_train, type_test
gc.collect()
if not debug:
np.save(str(result_filename), result_dict_lgb)
X_short.loc[X_short['type'] == type, 'oof'] = result_dict_lgb['oof']
X_short_test.loc[X_short_test['type'] == type, 'prediction'] = result_dict_lgb['prediction']
