dalian-university/秀水分区/data_features.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# time: 2023/4/12 17:42
# file: data_features.py
# author: David
# company: shenyang JY
import pandas as pd
from sklearn.model_selection import train_test_split
import numpy as np

class data_features(object):
    def __init__(self, opt):
        self.opt = opt
        self.time_step = self.opt.Model["time_step"]
        self.columns = list()

    def get_train_data(self, dfs):
        train_x, valid_x, train_y, valid_y = [], [], [], []
        for i, df in enumerate(dfs, start=1):
            datax, datay = self.get_data_features(df, is_train=True)
            tx, vx, ty, vy = train_test_split(datax, datay, test_size=self.opt.valid_data_rate, random_state=self.opt.Model["random_seed"], shuffle=self.opt.shuffle_train_data)  # 划分训练和验证集
            train_x.extend(tx)
            valid_x.extend(vx)
            train_y.extend(ty)
            valid_y.extend(vy)

        train_y = np.concatenate([[y.iloc[:, 1:].values for y in train_y]], axis=0)
        valid_y = np.concatenate([[y.iloc[:, 1:].values for y in valid_y]], axis=0)

        train_x = np.array([x[0].values for x in train_x])
        valid_x = np.array([x[0].values for x in valid_x])

        return train_x, valid_x, train_y, valid_y

    def get_test_data(self, dfs):
        test_x, test_y, data_y = [], [], []
        for i, df in enumerate(dfs, start=1):
            datax, datay = self.get_data_features(df, is_train=False)

            test_x.extend(datax)
            test_y.extend(datay)
            data_y.extend(datay)

        test_x = np.array([x[0].values for x in test_x])
        test_y = np.concatenate([[y.iloc[:, 1:].values for y in test_y]], axis=0)
        return test_x, test_y, data_y

    def get_data_features(self, norm_data, is_train):   # 这段代码基于pandas方法的优化
        time_step = self.opt.Model["time_step"]
        feature_data = norm_data.reset_index(drop=True)
        time_step_loc = time_step - 1
        train_num = int(len(feature_data))
        label_features = ['C_TIME', 'col1_power', 'col2_power', 'sum_power', 'C_VALUE'] if is_train is True else ['C_TIME', 'col1_power', 'col2_power', 'sum_power', 'C_VALUE']
        nwp = [feature_data.loc[i:i + time_step_loc, 'C_RADIATION':'C_TPR'].reset_index(drop=True) for i in range(train_num - time_step)]  # 数据库字段 'C_T': 'C_WS170'
        labels = [feature_data.loc[i:i + time_step_loc, label_features].reset_index(drop=True) for i in range(train_num - time_step)]
        features_x, features_y = [], []
        print("匹配环境前，{}组".format(len(nwp)), end=" -> ")
        for i, row in enumerate(zip(nwp, labels)):
            time_end = row[1]['C_TIME'][0]
            time_start = time_end - pd.DateOffset(1)
            # row1 = envir[(envir.C_TIME < time_end) & (envir.C_TIME > time_start)][-16:]
            # if len(row1) < 16:
            #     print("环境数据不足16个点：", len(row1))
            #     continue
            # row1 = row1.reset_index(drop=True).drop(['C_TIME'], axis=1)
            # features_x.append([row1.iloc[:,:-4], row1.iloc[:,-4:]])
            features_x.append([row[0]])
            features_y.append(row[1])
        print("匹配环境后，{}组".format(len(features_x)))
        return features_x, features_y