-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathgenerator.py
183 lines (157 loc) · 9.47 KB
/
generator.py
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
# coding: utf-8
import numpy as np
import csv
import pickle as P
import sys
class DataGenerator:
def __init__ (self) :
self.files = ["Data/2007.csv", "Data/2008.csv", "Data/2009.csv", "Data/2010.csv", "Data/2011.csv", "Data/2012.csv", "Data/2013.csv", "Data/2014.csv", "Data/2015.csv", "Data/2016.csv"]
self.target_result_name = ""
self.target_result_column = 14 # 着順（答え）
# self.target_race_colums = {1, 2, 4, 5, 6, 7, 8, 9} # 学習の要素
self.target_race_colums = {3, 4, 6, 7, 8, 9} # 学習の要素ーレース情報
# self.target_horse_colums = {11, 12, 14, 15, 17, 21, 22, 23, 24, 25} # 学習の要素ー馬の情報
self.target_horse_colums = {10, 11, 12, 13, 15, 16, 18, 19, 20, 21, 22, 24, 26} # 学習の要素ー馬の情報
# self.excluded_colums = {3, 13, 16, 18, 19, 26, 27, 28} # 学習には使わない要素
self.train_data_dimension = len(self.target_race_colums) + len(self.target_horse_colums) * 18 # 1レース18頭立てとして計算する
print("train_data_dimension = {}".format(self.train_data_dimension))
self.train_data = np.array([], dtype=np.float32).reshape(0, self.train_data_dimension)
self.train_data_answer = np.array([], dtype=np.int32)
# self.train_data_answer = np.array([], dtype=np.int32).reshape(0, 18)
# テスト対象
self.test_date = '0912'#'1610-中山' # この条件にマッチするレースを検証データとする '0712-阪神'#
self.test_data = np.array([], dtype=np.float32).reshape(0, self.train_data_dimension)
self.test_data_answer = np.array([], dtype=np.int32)
# self.test_data_answer = np.array([], dtype=np.int32).reshape(0, 18)
# 学習用に数値に変換する
self.dataMap = {
3 : { "札幌": 0, "函館": 1, "福島": 2, "東京": 3, "中山": 4, "京都": 5, "新潟": 6, "阪神": 7, "中京": 8, "小倉": 9 },
6 : { "芝" : 0, "ダ" : 1 },
9 : { "不" : 0, "重" : 1, "稍" : 2, "良" : 3 },
15 : {"牡" : 0, "牝" : 1, "セ" : 2},
18 : {"逃げ" : 0, "先行" : 1, "中団" : 2, "差し" : 3, "後方" : 4, "追込" : 5, "ﾏｸﾘ" : 6, "" : 7}
}
def read(self):
hurdle_race_count = 0
header = []
for file in self.files:
print("file name = {}".format(file))
with open(file, "r") as f:
reader = csv.reader(f)
previous_race_id = ""
race_parameter = np.array([], dtype=np.float32)
horses_parameter = np.array([], dtype=np.float32).reshape(0, len(self.target_horse_colums) + 1)
# 障害は除くデータで予測データを作成
answer = np.int32(-1)
for idx, row in enumerate(reader):
# if idx == 0: # skip header
# continue
# elif idx == 1: # skip header
# for i, col in enumerate(row):
# # if i in self.excluded_colums:
# # print("index = {}, name = {}, excluded from training".format(i, row[i]))
# if i in self.target_race_colums:
# print("index = {}, name = {}, used for training - race".format(i, row[i]))
# if i in self.target_horse_colums:
# print("index = {}, name = {}, used for training - horse".format(i, row[i]))
# header = row
# continue
# elif row[4] == '障害' :
# hurdle_race_count += 1
# continue
current_race_id = "{}{}{}-{}-{}".format(row[0], row[1], row[2], row[3], row[4]).replace(u'\ufeff', '')
current_date = "{}{}".format(row[0], row[1])
if len(previous_race_id) > 0 and previous_race_id != current_race_id:
sys.stdout.write("\rrace_id = {}".format(previous_race_id))
sys.stdout.flush()
# output previous race info
# print("race_parameter = {}".format(race_parameter))
# print("horses_parameter = {}".format(horses_parameter))
horses, race_answer = np.hsplit(horses_parameter, [len(self.target_horse_colums)]) # 馬の情報と着順を分離
rase_all_parameter = np.hstack((race_parameter, horses.flatten())) # レースと全馬の情報を結合
rase_all_parameter = np.pad(rase_all_parameter, (0, self.train_data_dimension - len(rase_all_parameter)), 'constant', constant_values=(0, -1)) # 頭数がすくない場合padding
race_answer = np.pad(race_answer.flatten(), (0, 18 - len(race_answer.flatten())), 'constant', constant_values=(0, 0)) # 頭数がすくない場合padding
race_answer = race_answer.astype(np.int32)
# print("race_answer = {}".format(race_answer))
# return
# print("loader.train_data = {}, horses_parameter = {}, rase_all_parameter = {}".format(self.train_data.dtype,horses_parameter.dtype, rase_all_parameter.dtype))
# print("loader.train_data_answer = {}".format(self.train_data_answer))
# print("loader.test_data = {}".format(self.test_data.dtype))
# print("loader.test_data_answer = {}".format(self.test_data_answer))
if current_date != self.test_date:
self.train_data = np.vstack([self.train_data, rase_all_parameter])
# self.train_data_answer = np.vstack([self.train_data_answer, race_answer])
# self.train_data_answer = np.vstack([self.train_data_answer, race_answer])
self.train_data_answer = np.append(self.train_data_answer, np.int32(answer))
else:
self.test_data = np.vstack([self.test_data, rase_all_parameter])
# self.test_data_answer = np.vstack([self.test_data_answer, race_answer])
self.test_data_answer = np.append(self.test_data_answer, np.int32(answer))
# initialize
horses_parameter = np.array([], dtype=np.float32).reshape(0, len(self.target_horse_colums) + 1)
else:
# print("current race = {}".format(current_race_id))
pass
# race_parameter_label = []
# horse_parameter_label = []
race_parameter = np.array([], dtype=np.float32)
horse_parameter = np.array([], dtype=np.float32)
# if current_race_id == '070811-札幌-8': # "080816-札幌-1": #070901-札幌-1":
# return
# マスタデータで数値化
for i, col in enumerate(row): # 馬一頭分の情報の詳細
# if i == 0:
# if self.test_row_no == -1 and col == self.test_date :
# self.test_row_no = (idx - hurdle_race_count)
# race_parameter_label.append(header[i])
# race_parameter = np.append(race_parameter, col.replace('-',''))
if i == self.target_result_column:
# self.target_result_name = header[i]
#answer = int(col) # 正解フラグを立てるならここをいじる。3以内の馬にフラグをたてる
# answer = 1 if int(col) == 1 else 0 # 正解フラグを立てるならここをいじる。1位の馬にフラグをたてる
if int(col) == 1:
answer = int(row[11]) - 1
# print("answer = {}".format(answer))
elif i in self.target_race_colums:
# race_parameter_label.append(header[i])
if i in self.dataMap :
race_parameter = np.append(race_parameter, np.float32(self.dataMap[i][col]))
else:
val = np.float32(col) if col else np.float32(0)
race_parameter = np.append(race_parameter, val)
elif i in self.target_horse_colums:
# horse_parameter_label.append(header[i])
if i in self.dataMap :
horse_parameter = np.append(horse_parameter, np.float32(self.dataMap[i][col]))
else:
val = np.float32(col) if col else np.float32(0) # 発走除外の馬などは空欄になる項目があるのでその場合は0で埋める
horse_parameter = np.append(horse_parameter, val)
if len(horse_parameter) == 0:
print("horse_parameter empty")
break
# print("horse_parameter = {}, answer = {}".format(horse_parameter.dtype, answer))
# return
horse_parameter = np.append(horse_parameter, np.float32(answer))
horses_parameter = np.vstack([horses_parameter, horse_parameter])
previous_race_id = current_race_id
def setup():
loader = DataGenerator()
loader.read()
with open('train_data.pickle', 'wb') as f:
P.dump(loader.train_data, f)
with open('train_data_answer.pickle', 'wb') as f:
P.dump(loader.train_data_answer, f)
with open('test_data.pickle', 'wb') as f:
P.dump(loader.test_data, f)
with open('test_data_answer.pickle', 'wb') as f:
P.dump(loader.test_data_answer, f)
print("train_data count = {}".format(len(loader.train_data)))
print("train_data_answer count = {}".format(len(loader.train_data_answer)))
print("test_data count = {}".format(len(loader.test_data)))
print("test_data_answer count = {}".format(len(loader.test_data_answer)))
print("loader.train_data = {}, shape = {}".format(loader.train_data.dtype, loader.train_data.shape))
print("loader.train_data_answer = {}, shape = {}".format(loader.train_data_answer.dtype, loader.train_data_answer.shape))
print("loader.test_data = {}, shape = {}".format(loader.test_data.dtype, loader.test_data.shape))
print("loader.test_data_answer = {}, shape = {}".format(loader.test_data_answer.dtype, loader.test_data_answer.shape))
if __name__ == '__main__':
setup()