def train_model(file_name, state, **kwds):
for key, value in kwds.items():
if key == 'cell_key':
cell_key = value
elif key == 'state':
state = value
elif key == 'pkl_dir':
pkl_dir = value[kwds['run_mode']]
elif key == 'config':
config = value[kwds['run_mode']]
elif key == 'processed_data_dir':
file_dir = value[kwds['run_mode']]
elif key == 'score_key':
score_key = value
data_x, data_y, cell_key = calc_feature_data(config, cell_key, score_key,
file_dir, file_name)
if data_x is None:
print('there is no data to train.')
return
ioo.save_data_csv(data_x.head(), 'columns_feature_' + file_name, pkl_dir)
mode = 'test'
res = build_model(data_x,
data_y,
split_mode=mode,
feature_method='f_regression',
pkl_dir=pkl_dir)
if mode == 'test':
d = {
'lr': '线性回归(LR)',
'dt': '决策树回归',
'rf': '随机森林',
'gbdt': 'GBDT',
'eva': '评估结果'
}
ioo.save_model_result(res, d, pkl_dir, state)
def save_distribute_data(data, filename, para_dict, mode, chunksize=None):
for col in para_dict['columns']:
if 'pca' in col:
data[col] = data[col].apply(str)
data[col] = data[col].apply(lambda x: x.replace('\n', ''))
data[col] = data[col].apply(lambda x: x.replace(',', ';'))
ioo.save_data_csv(data, filename, para_dict['data_dir'][mode])
def save_workstate_data(data, mask_filename, data_dir):
tmp = data[data['state'] == 1].reset_index(drop=True) #充电数据
filename = 'charge_' + mask_filename
ioo.save_data_csv(tmp, filename, data_dir)
tmp = data[data['state'] == 2].reset_index(drop=True) #放电数据
filename = 'discharge_' + mask_filename
ioo.save_data_csv(tmp, filename, data_dir)
print('the data has been save within each workstate.')
def save_workstate_data(regx, mask_filename, raw_data_dir, data_dir):
result, temp = gf.get_regx_data(regx, raw_data_dir)
if result:
import io_operation as ioo
data = temp[temp['current_mean'] == 0].reset_index(drop=True) #静置数据
filename = 'rest_' + mask_filename
ioo.save_data_csv(data, filename, data_dir)
data = temp[temp['current_mean'] > 0].reset_index(drop=True) #充电数据
filename = 'charge_' + mask_filename
ioo.save_data_csv(data, filename, data_dir)
data = temp[temp['current_mean'] < 0].reset_index(drop=True) #放电数据
filename = 'discharge_' + mask_filename
ioo.save_data_csv(data, filename, data_dir)
print('the data has been save within each workstate.')
return result
def save_bat_data(data, filename, para_dict, mode, chunksize=None):
ioo.save_data_csv(data, filename, para_dict['processed_data_dir'][mode])