TRMMA/dataset_spliter.py at main · coding-back01/TRMMA · 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
import argparse
import csv
import math
import os
from ast import literal_eval
import random
import numpy as np
import pandas as pd
import geopandas as gpd


def gen_path(filenames, zone, left=5, right=300, len_ratio=0.20):
    ratios = [0.1, 0.3, 0.5, 0.7, 0.9]
    print("[{}, {}), {}".format(left, right, len_ratio))
    data_rows = []
    for file in filenames:
        print("=====> {}".format(file))
        reader = csv.reader(open(file, "r"))
        cnt = 0
        cnt_filter = 0
        for item in reader:
            path = literal_eval(item[3])
            cnt += 1
            if float(item[8]) > len_ratio or len(path) < left or len(path) >= right:
                continue

            raw = literal_eval(item[4])
            a, b, _, c, d, *_ = zip(*raw)
            min_lon, max_lon, min_lat, max_lat = np.min(a + c), np.max(a + c), np.min(b + d), np.max(b + d)
            if min_lon <= zone[0] or max_lon >= zone[1] or min_lat <= zone[2] or max_lat >= zone[3]:
                continue

            idxs = []
            length = len(raw)
            for keep_ratio in ratios:
                low_sample = [0] + sorted(random.sample(range(1, length - 1), int((length - 2) * keep_ratio))) + [length - 1]
                idxs.append(low_sample)
            data_rows.append(item[:5] + [idxs])
            cnt_filter += 1
        print("# Input / Filtered Trajectories: {}, {}".format(cnt, cnt_filter))
    return data_rows


def get_zone(map_dir, dist):
    edges = gpd.read_file(os.path.join(map_dir, "edges.shp"))
    zone = [180, -180, 90, -90]
    for i in range(edges.shape[0]):
        tmp = edges.iloc[i]
        points = tmp['geometry'].coords
        zone[0] = min(zone[0], np.min(points.xy[0]))
        zone[1] = max(zone[1], np.max(points.xy[0]))
        zone[2] = min(zone[2], np.min(points.xy[1]))
        zone[3] = max(zone[3], np.max(points.xy[1]))
    print(zone)
    zone_s = [zone[0] + dist,
              zone[1] - dist,
              zone[2] + dist,
              zone[3] - dist]
    print(zone_s)
    return zone_s


def get_stats(filenames):
    data_rows = []
    cnt = 0
    for file in filenames:
        print("=====> {}".format(file))
        reader = csv.reader(open(file, "r"))
        for item in reader:
            cnt += 1
            raw = literal_eval(item[4])
            data_rows.append([item[6], item[7], item[8], item[9], item[10], item[11], item[5], len(raw)])
    print(cnt, len(data_rows))
    return data_rows


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--data_dir', type=str, default="../data/chengdu_NMLR")
    parser.add_argument('--workspace', type=str, default="../data/len_test")
    parser.add_argument('--map_dir', type=str, default="sf")
    parser.add_argument('-ratio', type=str, default="0.6,0.2,0.2")
    parser.add_argument('-group_num', type=int, default=5)
    parser.add_argument('-quality', type=float, default=0.20)
    parser.add_argument('-mode', type=int, default=1)
    parser.add_argument('-left', type=int, default=5)
    parser.add_argument('-right', type=int, default=300)
    parser.add_argument('-small', type=int, default=-1)
    parser.add_argument('-time', action="store_true", default=False)
    parser.add_argument('-dist', type=float, default=0.0005)
    args = parser.parse_args()
    print(args)

    files = []
    if args.mode in [0, 1]:
        for item in os.listdir(args.data_dir):
            path = os.path.join(args.data_dir, item)
            if os.path.isfile(path):
                files.append(path)
    print("files number: {}".format(len(files)))

    if args.mode == 0:
        stats = get_stats(files)
        with open(os.path.join(args.workspace, "matching_ratio.csv"), 'w') as fp:
            fields_output_file = csv.writer(fp, delimiter=',')
            fields_output_file.writerows(stats)
    elif args.mode == 1:
        zone = get_zone(args.map_dir, args.dist)
        paths = gen_path(files, zone, len_ratio=args.quality, left=args.left, right=args.right)
        print("==> Number of Trajectories: {}".format(len(paths)))

        if args.small > 0:
            paths = random.sample(paths, args.small)

        # shuffle data
        ratio = args.ratio.split(",")
        ratio = [float(item) for item in ratio]
        print("Input ratio: {}".format(ratio))
        ratio = [r / sum(ratio) for r in ratio]
        print("Normalized ratio: {}".format(ratio))

        data_size = len(paths)
        print("==> Trajectories data size: {}".format(data_size))
        if args.time:
            time_order = []
            for i, item in enumerate(paths):
                tmp = literal_eval(item[2])
                time_order.append([i, tmp[0][1]])
            time_order.sort(key=lambda elem: elem[1])
            index, _ = zip(*time_order)
            index = list(index)
        else:
            index = [i for i in range(data_size)]
            random.shuffle(index)
        paths = np.array(paths, dtype=object)
        train = paths[index[0: int(data_size * ratio[0])]].tolist()
        valid = paths[index[int(data_size * ratio[0]): int(data_size * (ratio[0] + ratio[1]))]].tolist()
        test = paths[index[int(data_size * (ratio[0] + ratio[1])): data_size]].tolist()

        num = args.group_num
        scale = math.ceil(len(train) * 1.0 / num)
        label = int(100 / num)
        print("# Groups: {}, Scale: {}".format(num, scale))
        for i in range(num):
            with open(os.path.join(args.workspace, "traj_train_{}".format(label * (i+1))), 'w') as fp:
                fields_output_file = csv.writer(fp, delimiter=',')
                fields_output_file.writerows(train[:scale * (i+1)])

        with open(os.path.join(args.workspace, "traj_valid"), 'w') as fp:
            fields_output_file = csv.writer(fp, delimiter=',')
            fields_output_file.writerows(valid)
        with open(os.path.join(args.workspace, "traj_test"), 'w') as fp:
            fields_output_file = csv.writer(fp, delimiter=',')
            fields_output_file.writerows(test)
        print("Training: {}, Validation: {}, Test: {}".format(len(train), len(valid), len(test)))
    elif args.mode == 2:
        train = pd.read_csv(os.path.join(args.workspace, 'traj_train_20'), sep=",", header=None, names=['oid', 'tid', 'offsets', 'path', 'raw', 'low']).to_numpy()
        num = train.shape[0] // 20
        scale = 1
        groups = [1, 3, 5, 10]
        for item in groups:
            with open(os.path.join(args.workspace, "traj_train_{}".format(scale * item)), 'w') as fp:
                fields_output_file = csv.writer(fp, delimiter=',')
                fields_output_file.writerows(train[:num * item])