質問編集履歴
2
文法修正
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -283,3 +283,39 @@
|
|
|
283
283
|
main()
|
|
284
284
|
|
|
285
285
|
```
|
|
286
|
+
|
|
287
|
+
|
|
288
|
+
|
|
289
|
+
```ここに言語を入力
|
|
290
|
+
|
|
291
|
+
Ypred [1 1 1 ... 6 6 6]
|
|
292
|
+
|
|
293
|
+
|
|
294
|
+
|
|
295
|
+
Ytest
|
|
296
|
+
|
|
297
|
+
384 1
|
|
298
|
+
|
|
299
|
+
385 1
|
|
300
|
+
|
|
301
|
+
386 1
|
|
302
|
+
|
|
303
|
+
387 1
|
|
304
|
+
|
|
305
|
+
388 1
|
|
306
|
+
|
|
307
|
+
..
|
|
308
|
+
|
|
309
|
+
7675 8
|
|
310
|
+
|
|
311
|
+
7676 8
|
|
312
|
+
|
|
313
|
+
7677 8
|
|
314
|
+
|
|
315
|
+
7678 8
|
|
316
|
+
|
|
317
|
+
7679 8
|
|
318
|
+
|
|
319
|
+
Name: category, Length: 1536, dtype: in
|
|
320
|
+
|
|
321
|
+
```
|
1
改善
test
CHANGED
|
File without changes
|
test
CHANGED
|
@@ -49,3 +49,237 @@
|
|
|
49
49
|
SyntaxError: invalid character in identifier
|
|
50
50
|
|
|
51
51
|
```
|
|
52
|
+
|
|
53
|
+
|
|
54
|
+
|
|
55
|
+
コード本文です。
|
|
56
|
+
|
|
57
|
+
```ここに言語を入力
|
|
58
|
+
|
|
59
|
+
# -*- coding: utf-8 -*-
|
|
60
|
+
|
|
61
|
+
|
|
62
|
+
|
|
63
|
+
import scipy as sp
|
|
64
|
+
|
|
65
|
+
import pandas as pd
|
|
66
|
+
|
|
67
|
+
from pandas import Series, DataFrame
|
|
68
|
+
|
|
69
|
+
import matplotlib.pyplot as plt
|
|
70
|
+
|
|
71
|
+
from sklearn.preprocessing import StandardScaler
|
|
72
|
+
|
|
73
|
+
from sklearn.ensemble import RandomForestClassifier #RandomForest
|
|
74
|
+
|
|
75
|
+
from sklearn.svm import SVC # SVM用
|
|
76
|
+
|
|
77
|
+
import lightgbm as lgb
|
|
78
|
+
|
|
79
|
+
from sklearn.metrics import confusion_matrix #混同行列
|
|
80
|
+
|
|
81
|
+
from sklearn.metrics import accuracy_score, precision_score #適合率
|
|
82
|
+
|
|
83
|
+
from sklearn.metrics import recall_score, f1_score #再現率,F1スコア
|
|
84
|
+
|
|
85
|
+
from sklearn.metrics import make_scorer
|
|
86
|
+
|
|
87
|
+
|
|
88
|
+
|
|
89
|
+
|
|
90
|
+
|
|
91
|
+
|
|
92
|
+
|
|
93
|
+
|
|
94
|
+
|
|
95
|
+
|
|
96
|
+
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
|
|
100
|
+
|
|
101
|
+
def SVM(X_train_std, X_test_std, Y_train, Y_test):
|
|
102
|
+
|
|
103
|
+
model_SVM = SVC(random_state=0, kernel = "rbf", C = 1000 , gamma = 0.01 )
|
|
104
|
+
|
|
105
|
+
#学習モデル構築。引数に訓練データの特徴量と、それに対応したラベル
|
|
106
|
+
|
|
107
|
+
model_SVM.fit(X_train_std,Y_train)
|
|
108
|
+
|
|
109
|
+
|
|
110
|
+
|
|
111
|
+
#予測したクラスラベル
|
|
112
|
+
|
|
113
|
+
Y_pred = model_SVM.predict(X_test_std)
|
|
114
|
+
|
|
115
|
+
|
|
116
|
+
|
|
117
|
+
# .scoreで正解率を算出。
|
|
118
|
+
|
|
119
|
+
print("\nSVM")
|
|
120
|
+
|
|
121
|
+
print("train score:",model_SVM.score(X_train_std,Y_train))
|
|
122
|
+
|
|
123
|
+
print("test score:",model_SVM.score(X_test_std,Y_test))
|
|
124
|
+
|
|
125
|
+
|
|
126
|
+
|
|
127
|
+
print("accuracy score:",accuracy_score(Y_test, Y_pred,average='micro'))
|
|
128
|
+
|
|
129
|
+
print("precision score:",precision_score(Y_test, Y_pred,average='micro'))
|
|
130
|
+
|
|
131
|
+
print("recall score:",recall_score(Y_test, Y_pred,average='micro'))
|
|
132
|
+
|
|
133
|
+
print("f1 score:",f1_score(Y_test, Y_pred,average='micro'))
|
|
134
|
+
|
|
135
|
+
|
|
136
|
+
|
|
137
|
+
|
|
138
|
+
|
|
139
|
+
|
|
140
|
+
|
|
141
|
+
def GBM(X_train, X_test, Y_train, Y_test):
|
|
142
|
+
|
|
143
|
+
model_GBM = lgb.LGBMClassifier(boosting_type='gbdt', num_leaves=58,
|
|
144
|
+
|
|
145
|
+
max_depth=14, learning_rate=0.1, n_estimators=940,
|
|
146
|
+
|
|
147
|
+
min_child_samples=40, importance_type="split", random_state=0)
|
|
148
|
+
|
|
149
|
+
|
|
150
|
+
|
|
151
|
+
#学習モデル構築。引数に訓練データの特徴量と、それに対応したラベル
|
|
152
|
+
|
|
153
|
+
model_GBM.fit(X_train, Y_train)
|
|
154
|
+
|
|
155
|
+
|
|
156
|
+
|
|
157
|
+
#予測したクラスラベル
|
|
158
|
+
|
|
159
|
+
Y_pred = model_GBM.predict(X_test)
|
|
160
|
+
|
|
161
|
+
|
|
162
|
+
|
|
163
|
+
# .scoreで正解率を算出。
|
|
164
|
+
|
|
165
|
+
print("\nGBM")
|
|
166
|
+
|
|
167
|
+
print("train score:",model_GBM.score(X_train,Y_train))
|
|
168
|
+
|
|
169
|
+
print("test score:",model_GBM.score(X_test,Y_test))
|
|
170
|
+
|
|
171
|
+
|
|
172
|
+
|
|
173
|
+
print("accuracy score:",accuracy_score(Y_test, Y_pred,average='micro'))
|
|
174
|
+
|
|
175
|
+
print("precision score:",precision_score(Y_test, Y_pred,average='micro'))
|
|
176
|
+
|
|
177
|
+
print("recall score:",recall_score(Y_test, Y_pred,average='micro'))
|
|
178
|
+
|
|
179
|
+
print("f1 score:",f1_score(Y_test, Y_pred,average='micro'))
|
|
180
|
+
|
|
181
|
+
|
|
182
|
+
|
|
183
|
+
|
|
184
|
+
|
|
185
|
+
|
|
186
|
+
|
|
187
|
+
|
|
188
|
+
|
|
189
|
+
|
|
190
|
+
|
|
191
|
+
|
|
192
|
+
|
|
193
|
+
|
|
194
|
+
|
|
195
|
+
|
|
196
|
+
|
|
197
|
+
def main():
|
|
198
|
+
|
|
199
|
+
"""
|
|
200
|
+
|
|
201
|
+
TODO case_nameに任意の名前を指定
|
|
202
|
+
|
|
203
|
+
フォルダを統一するために以降のscriptも名前を統一する
|
|
204
|
+
|
|
205
|
+
"""
|
|
206
|
+
|
|
207
|
+
case_name = "case3"
|
|
208
|
+
|
|
209
|
+
# ---------------------------------------
|
|
210
|
+
|
|
211
|
+
|
|
212
|
+
|
|
213
|
+
case_dir = "./casestudy/{}/".format(case_name)
|
|
214
|
+
|
|
215
|
+
input_csv_name = "2_extracted_features_original.csv"
|
|
216
|
+
|
|
217
|
+
input_csv_path = case_dir + input_csv_name
|
|
218
|
+
|
|
219
|
+
input_df = pd.read_csv(input_csv_path, encoding="utf-8-sig")
|
|
220
|
+
|
|
221
|
+
|
|
222
|
+
|
|
223
|
+
|
|
224
|
+
|
|
225
|
+
#すべてのデータを対象に分類を行う場合
|
|
226
|
+
|
|
227
|
+
#-----------------------------------------------------------------------
|
|
228
|
+
|
|
229
|
+
task = "all"
|
|
230
|
+
|
|
231
|
+
train_df = input_df[input_df["train_test_flag"] == 0]
|
|
232
|
+
|
|
233
|
+
test_df = input_df[input_df["train_test_flag"] == 1]
|
|
234
|
+
|
|
235
|
+
# print(train_df)
|
|
236
|
+
|
|
237
|
+
# print(test_df)
|
|
238
|
+
|
|
239
|
+
|
|
240
|
+
|
|
241
|
+
X_train = train_df.loc[:, "contrast":"inverse_difference_m_norm"]
|
|
242
|
+
|
|
243
|
+
X_test = test_df.loc[:, "contrast":"inverse_difference_m_norm"]
|
|
244
|
+
|
|
245
|
+
# print(X_train)
|
|
246
|
+
|
|
247
|
+
# print(X_test)
|
|
248
|
+
|
|
249
|
+
Y_train = train_df["category"]
|
|
250
|
+
|
|
251
|
+
Y_test = test_df["category"]
|
|
252
|
+
|
|
253
|
+
# print(Y_train)
|
|
254
|
+
|
|
255
|
+
# print(Y_test)
|
|
256
|
+
|
|
257
|
+
|
|
258
|
+
|
|
259
|
+
sc = StandardScaler()
|
|
260
|
+
|
|
261
|
+
sc.fit(X_train)
|
|
262
|
+
|
|
263
|
+
X_train_std = sc.transform(X_train)
|
|
264
|
+
|
|
265
|
+
X_test_std = sc.transform(X_test)
|
|
266
|
+
|
|
267
|
+
# print(X_train_std)
|
|
268
|
+
|
|
269
|
+
# print(X_test_std)
|
|
270
|
+
|
|
271
|
+
# #------------------------------------------------------------------------
|
|
272
|
+
|
|
273
|
+
SVM(X_train_std, X_test_std, Y_train, Y_test)
|
|
274
|
+
|
|
275
|
+
GBM(X_train, X_test, Y_train, Y_test)
|
|
276
|
+
|
|
277
|
+
|
|
278
|
+
|
|
279
|
+
|
|
280
|
+
|
|
281
|
+
if __name__ == "__main__":
|
|
282
|
+
|
|
283
|
+
main()
|
|
284
|
+
|
|
285
|
+
```
|