best_config = {} predict_days = list(range(1, 6)) #The future # day wish model to predict consider_lagdays = list(range( 1, 6)) #Contain # lagday information for a training input model_name = 'xgb_2cls' config = mc.model_config(model_name).get #srcPath = '/home/ubuntu/dataset/etf_prediction/0601/all_feature_data_Nm_1_MinMax_120.pkl' #metaPath = '/home/ubuntu/dataset/etf_prediction/0601/all_meta_data_Nm_1_MinMax_120.pkl' #corrDate_path = '/home/ubuntu/dataset/etf_prediction/0601/xcorr_date_data.pkl' srcPath = '../../Data/0601/all_feature_data_Nm_1_MinMax_120.pkl' metaPath = '../../Data/0601/all_meta_data_Nm_1_MinMax_120.pkl' corrDate_path = '../../Data/0601/xcorr_date_data.pkl' *_, meta = gu.read_metafile(metaPath) corrDate = gu.read_datefile(corrDate_path) corrDate_range = list(range(3, len(corrDate['0050']), 3)) tv_gen = dp.train_validation_generaotr() f = tv_gen._load_data(srcPath) total_progress = len(stock_list) * len(predict_days) * len( consider_lagdays) * len(feature_list_comb_noraml) * len(corrDate_range) progress = tqdm(total=total_progress) progress.set_description("[SP][{}]".format(model_name)) for s in stock_list: best_config[s] = {} progress.set_description("[SP][{}][{}]".format(model_name, s)) #if s == '0050': _stock_list = ['0050', '2330'] #else: _stock_list = [s]
] stock_list = ['00690'] best_config = {} predict_days = list(range(1, 6)) #The future # day wish model to predict consider_lagdays = list(range( 1, 6)) #Contain # lagday information for a training input feature_list_comb = [['velocity'], ['ma'], ['ratio'], ['rsi'], ['kdj'], ['macd'], ['ud']] config = mc.model_config('stack').get srcPath = '/home/ubuntu/dataset/etf_prediction/all_feature_data_Nm_1_MinMax_94.pkl' metaPath = '/home/ubuntu/dataset/etf_prediction/all_meta_data_Nm_1_MinMax_94.pkl' *_, meta = gu.read_metafile(metaPath) corrDate = gu.read_datefile( '/home/ubuntu/dataset/etf_prediction/corr_date/xcorr_date_data.pkl') corrDate_range = list(range(3, len(corrDate['0050']) + 1)) tv_gen = dp.train_validation_generaotr() f = tv_gen._load_data(srcPath) for s in stock_list: best_config[s] = {} for predict_day in predict_days: best_config[s][predict_day] = {} best_accuracy = 0 best_test_accuracy = 0 for consider_lagday in consider_lagdays: for feature_list in feature_list_comb: