def main():
pre_df = csv_to_df("./dataset/", "data.csv")
feature_matrix = feature_eng(pre_df)
label_matrix = label_eng(pre_df)
generator = cv_generator(feature_matrix, label_matrix)
#running_model = model_LR()
#running_model = model_SVM()
running_model = model_XGBoost()
frame_classification(generator, running_model, feature_matrix,
label_matrix)
def main():
pre_df = csv_to_df("./dataset/", "data.csv")
feature, label, feature_mean, feature_std = op_feature_ext(pre_df, "Rs")
running_model = model_XGBoost()
input_sample, feature_df, origin_col = op_sample_reader(
"./dataset/", "opt_data.csv", "Rs", feature_mean, feature_std)
op_result = para_optimize(running_model, feature, label, feature_mean,
feature_std, input_sample)
print(input_sample)
print(op_result)
def op_sample_reader(sample_path, sample_name, col_name, pred_model,
feature_mean, feature_std):
pre_df = csv_to_df(sample_path, sample_name, pred_model)
origin_col = pre_df[col_name]
feature_df = pre_df.drop(columns=[col_name], axis=1)
pre_feature = feature_df.values
raw_feature = pre_feature[:, 0:-1]
raw_feature_normalized = (raw_feature - feature_mean) / feature_std
return raw_feature_normalized, pre_df, origin_col
def pred_sample_reader(sample_path, sample_name, pred_model, feature_mean,
feature_std):
pre_df = csv_to_df(sample_path, sample_name, pred_model)
pro_df = pre_df.drop(columns=["label"], axis=1)
pro_df['ApxRs'] = pro_df['Ap'] * pro_df['Rs']
pro_df['AexRs'] = pro_df['Ae'] * pro_df['Rs']
pro_df['AexAp'] = pro_df['Ae'] * pro_df['Ap']
pro_df['ApxRsxAe'] = pro_df['Ap'] * pro_df['Rs'] * pro_df['Ae']
pro_feature = pro_df.values
pro_feature_normalized = (pro_feature - feature_mean) / feature_std
return pro_feature_normalized, pre_df
def main():
raw_df = csv_to_df("./dataset/", "data.csv")
columns_entropy = [(col, calcu_each_gain(raw_df[col], raw_df))
for col in raw_df.iloc[:, :-1]]
print(columns_entropy)
def main(args):
print(" ****** Running function list ****** ")
if args.run_infogain is True:
print("Priority of parameters influencing product quality")
if args.run_qualitypredict is True:
print("Product quality predict")
if args.run_parameteroptimize is True:
print("Process parameter optimize")
print(" ****** Loading trianing data from %s ******"%(args.file_path + args.file_name))
raw_df = csv_to_df(args.file_path, args.file_name, args.pred_model)
print(" *** Head of original dataset *** ")
print(raw_df.head())
feature_matrix, ori_feature_mean, ori_feature_std = feature_eng(raw_df)
label_matrix = label_eng(raw_df)
if args.run_infogain is True:
print(" ****** Priority of parameters influencing product quality ****** ")
columns_entropy = [(col, info_gain.calcu_each_gain(raw_df[col], raw_df)) for col in raw_df.iloc[:, :-1]]
print(" *** Information Gain of process parameter *** ")
print(columns_entropy)
if args.run_qualitypredict is True:
print(" ****** Product quality predict ****** ")
if args.pred_model == "XGBoost":
generator = quality_predict.cv_generator(feature_matrix, label_matrix)
else:
accuracy_label_list = []
for ele in label_matrix:
if ele >= 1.62:
accuracy_label_list.append(1)
else:
accuracy_label_list.append(0)
label_matrix_int = np.array(accuracy_label_list)
generator = quality_predict.cv_generator(feature_matrix, label_matrix_int)
print(" *** Training process *** ")
if args.pred_model == "XGBoost":
running_model = quality_predict.model_XGBoost()
quality_predict.frame_classification(generator,
running_model,
feature_matrix,
label_matrix,
args.output_dir)
else:
running_model = quality_predict.model_GBDT()
quality_predict.frame_regression(generator,
running_model,
feature_matrix,
label_matrix,
args.output_dir)
pred_norm_feature, pre_pred_df = quality_predict.pred_sample_reader(args.file_path,
args.pred_file_name,
args.pred_model,
ori_feature_mean,
ori_feature_std)
pred_model_dir = args.output_dir + "model_1"
pred_result = quality_predict.qual_pred(args.pred_model,
pred_model_dir,
pred_norm_feature)
print(" *** Product quality predict *** ")
print("Origin process parameter")
print(pre_pred_df)
print("Quality result prediction ")
print(pred_result)
if args.run_parameteroptimize is True:
print(" ****** Process parameter optimize ****** ")
feature, label, feature_mean, feature_std = op_feature_ext(raw_df, args.opt_parameter)
if args.pred_model == "XGBoost":
running_model = quality_predict.model_XGBoost()
else:
running_model = quality_predict.model_GBDT()
input_sample, ori_fea, ori_col = op_sample_reader(args.file_path,
args.opt_file_name,
args.opt_parameter,
args.pred_model,
feature_mean,
feature_std)
op_result = para_optimize(running_model,
feature,
label,
args.pred_model,
input_sample)
print(" *** Process parameter optimize *** ")
print("Origin process parameter")
print(ori_fea)
print("Optimized %s"%(args.opt_parameter))
print(op_result)