Simple_MLP/dataset.py at main · EnzeXu/Simple_MLP · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
from torch.utils.data import Dataset, DataLoader
# from const import *
# from sklearn.preprocessing import MinMaxScaler
import numpy as np
import os
import pandas as pd

import pickle
import random
import time
import torch

from torch.utils.data import Dataset, DataLoader
from sklearn.model_selection import train_test_split


from utils import my_min_max, decode

class MyDataset(Dataset):
    def __init__(self, path, filter):
        print(f"loading data from {path} ...")
        df = pd.read_csv(path, skiprows=1)
        if df.values.shape[-1] != 11:
            df = pd.read_csv(path, header=None)

        n_col = df.values.shape[-1]
        assert n_col in [6, 7, 11]

        record = dict()

        if n_col == 7:  # for 3->1
            self.x_data = torch.tensor(df.values, dtype=torch.float64)[:, :3]
            self.y_data = torch.tensor(df.values, dtype=torch.float64)[:, 3:4]
            self.x_data[:, 0:1], x1_min, x1_max = my_min_max(self.x_data[:, 0:1])
            self.x_data[:, 1:2], x2_min, x2_max = my_min_max(self.x_data[:, 1:2])
            self.x_data[:, 2:3], x3_min, x3_max = my_min_max(self.x_data[:, 2:3])
            record["x1_min"] = x1_min
            record["x1_max"] = x1_max
            record["x2_min"] = x2_min
            record["x2_max"] = x2_max
            record["x3_min"] = x3_min
            record["x3_max"] = x3_max
        elif n_col == 6:  # for 2->1
            self.x_data = torch.tensor(df.values, dtype=torch.float64)[:, :2]
            self.y_data = torch.tensor(df.values, dtype=torch.float64)[:, 2:3]
            self.x_data[:, 0:1], x1_min, x1_max = my_min_max(self.x_data[:, 0:1])
            self.x_data[:, 1:2], x2_min, x2_max = my_min_max(self.x_data[:, 1:2])
            record["x1_min"] = x1_min
            record["x1_max"] = x1_max
            record["x2_min"] = x2_min
            record["x2_max"] = x2_max
        else:  # for 7->1
            self.x_data = torch.tensor(df.values, dtype=torch.float64)[:, :7]
            self.y_data = torch.tensor(df.values, dtype=torch.float64)[:, 7:8]
            self.x_data[:, 0:1], x1_min, x1_max = my_min_max(self.x_data[:, 0:1])
            self.x_data[:, 1:2], x2_min, x2_max = my_min_max(self.x_data[:, 1:2])
            self.x_data[:, 2:3], x3_min, x3_max = my_min_max(self.x_data[:, 2:3])
            self.x_data[:, 3:4], x4_min, x4_max = my_min_max(self.x_data[:, 3:4])
            self.x_data[:, 4:5], x5_min, x5_max = my_min_max(self.x_data[:, 4:5])
            self.x_data[:, 5:6], x6_min, x6_max = my_min_max(self.x_data[:, 5:6])
            self.x_data[:, 6:7], x7_min, x7_max = my_min_max(self.x_data[:, 6:7])
            record["x1_min"] = x1_min
            record["x1_max"] = x1_max
            record["x2_min"] = x2_min
            record["x2_max"] = x2_max
            record["x3_min"] = x3_min
            record["x3_max"] = x3_max
            record["x4_min"] = x4_min
            record["x4_max"] = x4_max
            record["x5_min"] = x5_min
            record["x5_max"] = x5_max
            record["x6_min"] = x6_min
            record["x6_max"] = x6_max
            record["x7_min"] = x7_min
            record["x7_max"] = x7_max


        print(f"In generating dataset, n_col = {n_col}, and keys of record: {list(record.keys())}")

        self.y_data, y_min, y_max = my_min_max(self.y_data)

        record["y_min"] = y_min
        record["y_max"] = y_max
        with open(f"processed/filter={filter}/record_min_max.pkl", "wb") as f:
            pickle.dump(record, f)
        self.x_dim = self.x_data.shape[-1]
        self.y_dim = self.y_data.shape[-1]
        print(f"Full x shape: {self.x_data.shape}")
        print(f"Full y shape: {self.y_data.shape}")

    def __len__(self):
        return len(self.x_data)

    def __getitem__(self, idx):
        x = self.x_data[idx]
        y = self.y_data[idx]
        return x, y


# def one_time_generate_dataset():
#     t0 = time.time()
#     dataset = MyDataset("data/dataset_osci_0_1_2_v0604.csv")
#
#     print(dataset.x_data[0], dataset.y_data[0])
#
#     # train_idxs, val_idxs, test_idxs = train_test_split(np.arange(len(dataset)), test_size=0.2, random_state=0)
#     train_idxs, val_idxs = train_test_split(np.arange(len(dataset)), test_size=0.2, random_state=42)
#     print(len(train_idxs), len(val_idxs))
#     train_dataset = torch.utils.data.Subset(dataset, train_idxs)
#     val_dataset = torch.utils.data.Subset(dataset, val_idxs)
#     with open("processed/train_idx.pkl", "wb") as f:
#         pickle.dump(train_idxs, f)
#     with open("processed/val_idx.pkl", "wb") as f:
#         pickle.dump(val_idxs, f)
#
#     with open("processed/x_raw.pkl", "wb") as f:
#         pickle.dump(dataset.x_data, f)
#     with open("processed/y_raw.pkl", "wb") as f:
#         pickle.dump(dataset.y_data, f)
#
#     with open("processed/all.pkl", "wb") as f:
#         pickle.dump(dataset, f)
#     with open("processed/train.pkl", "wb") as f:
#         pickle.dump(train_dataset, f)
#     with open("processed/valid.pkl", "wb") as f:
#         pickle.dump(val_dataset, f)
#     print("cost {0:.6f} min".format((time.time() - t0) / 60.0))

def one_time_generate_dataset(source_path, filter):
    t0 = time.time()
    source_path = source_path.replace(".csv", f"_{filter}.csv")
    save_folder_path = f"processed/filter={filter}/"
    if not os.path.exists(save_folder_path):
        os.makedirs(save_folder_path)
    dataset = MyDataset(source_path, filter)


    print(dataset.x_data[0], dataset.y_data[0])

    # train_idxs, val_idxs, test_idxs = train_test_split(np.arange(len(dataset)), test_size=0.2, random_state=0)
    train_idxs, val_idxs = train_test_split(np.arange(len(dataset)), test_size=0.2, random_state=42)
    print(len(train_idxs), len(val_idxs))
    train_dataset = torch.utils.data.Subset(dataset, train_idxs)
    val_dataset = torch.utils.data.Subset(dataset, val_idxs)
    with open(save_folder_path + "train_idx.pkl", "wb") as f:
        pickle.dump(train_idxs, f)
    with open(save_folder_path + "val_idx.pkl", "wb") as f:
        pickle.dump(val_idxs, f)

    with open(save_folder_path + "x_raw.pkl", "wb") as f:
        pickle.dump(dataset.x_data, f)
    with open(save_folder_path + "y_raw.pkl", "wb") as f:
        pickle.dump(dataset.y_data, f)

    with open(save_folder_path + "all.pkl", "wb") as f:
        pickle.dump(dataset, f)
    with open(save_folder_path + "train.pkl", "wb") as f:
        pickle.dump(train_dataset, f)
    with open(save_folder_path + "valid.pkl", "wb") as f:
        pickle.dump(val_dataset, f)
    print(f"saved to {save_folder_path}")
    print("cost {0:.6f} min".format((time.time() - t0) / 60.0))


if __name__ == "__main__":
    # one_time_generate_dataset("data/dataset_osci_0_1_2_v0604.csv", "all")
    # one_time_generate_dataset("data/dataset_osci_0_1_2_v0604.csv", "200")
    # one_time_generate_dataset("data/dataset_osci_0_1_2_v0604.csv", "100")

    # one_time_generate_dataset("data/dataset_osci_3_4_5_v0604.csv", "all")
    # one_time_generate_dataset("data/dataset_osci_3_4_5_v0604.csv", "200")
    # one_time_generate_dataset("data/dataset_osci_3_4_5_v0604.csv", "100")

    # one_time_generate_dataset("data/dataset_osci_0_1_v0618.csv", "all")
    # one_time_generate_dataset("data/dataset_osci_0_1_v0618.csv", "200")
    # one_time_generate_dataset("data/dataset_osci_0_1_v0618.csv", "100")

    # one_time_generate_dataset("data/dataset_osci_v0628_large.csv", "all")
    # one_time_generate_dataset("data/dataset_osci_v0628_large.csv", "200")
    # one_time_generate_dataset("data/dataset_osci_v0628_large.csv", "100")

    one_time_generate_dataset("data/dataset_osci_v0628_small.csv", "all")
    one_time_generate_dataset("data/dataset_osci_v0628_small.csv", "200")
    one_time_generate_dataset("data/dataset_osci_v0628_small.csv", "100")