teratail
回答率
85.48%
質問するログイン新規登録
トップに関する質問mnistデータで最適なkを探すプログラム

編集履歴

質問編集履歴

6

clf = svm.SVC() clf.fit(train_data, label_train) pre = clf.predict()　を削除

2017/11/28 09:07

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -16,25 +16,27 @@
16
16
 
17
17
 
18
18
 
19
- ・TypeError Traceback (most recent call last)
19
+ ・NameError Traceback (most recent call last)
20
-
20
+
21
- <ipython-input-20-71497c22a695> in <module>()
21
+ <ipython-input-24-b95cedfc1745> in <module>()
22
-
23
- 4 clf = svm.SVC()
22
+
24
-
25
- 5 clf.fit(train_data, label_train)
23
+ 28 print('k={0}: {1}'.format(k, ac_score))
24
+
26
-
25
+ 29
26
+
27
- ----> 6 pre = clf.predict()
27
+ ---> 30 accuracy_scores.append(score)
28
-
28
+
29
- 7
29
+ 31
30
-
30
+
31
- 8 # 使う近傍数ごとに正解率＆各経過時間を計算
31
+ 32 # 各経過時間を表示
32
-
33
-
34
-
32
+
33
+
34
+
35
- TypeError: predict() missing 1 required positional argument: 'X'
35
+ NameError: name 'score' is not defined
36
+
37
+
38
+
36
-
39
+ ・Digitsデータではそのようなエラーはなかった
37
-
38
40
 
39
41
 
40
42
 
@@ -158,18 +160,14 @@
158
160
 
159
161
  train_data, test_data, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
160
162
 
163
+
164
+
161
165
  K = 10
162
166
 
163
167
  ks = range(1, K + 1)
164
168
 
165
169
 
166
170
 
167
- clf = svm.SVC()
168
-
169
- clf.fit(train_data, label_train)
170
-
171
- pre = clf.predict()
172
-
173
171
 
174
172
 
175
173
  # 使う近傍数ごとに正解率＆各経過時間を計算
@@ -208,7 +206,7 @@
208
206
 
209
207
  # 一つだけ取り除いたテストデータを識別
210
208
 
211
- predicted_label = model.predict(features[test])
209
+ predicted_label = model.predict(train_data[test])
212
210
 
213
211
  predicted_labels.append(predicted_label)
214
212
 

5

修正しました。

2017/11/28 09:07

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -16,6 +16,28 @@
16
16
 
17
17
 
18
18
 
19
+ ・TypeError Traceback (most recent call last)
20
+
21
+ <ipython-input-20-71497c22a695> in <module>()
22
+
23
+ 4 clf = svm.SVC()
24
+
25
+ 5 clf.fit(train_data, label_train)
26
+
27
+ ----> 6 pre = clf.predict()
28
+
29
+ 7
30
+
31
+ 8 # 使う近傍数ごとに正解率＆各経過時間を計算
32
+
33
+
34
+
35
+ TypeError: predict() missing 1 required positional argument: 'X'
36
+
37
+
38
+
39
+
40
+
19
41
 
20
42
 
21
43
  ###Digitsデータで最適なkを探すknnのプログラム
@@ -134,9 +156,7 @@
134
156
 
135
157
  test_size = 100
136
158
 
137
- train_data, test_data_, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
159
+ train_data, test_data, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
138
-
139
-
140
160
 
141
161
  K = 10
142
162
 
@@ -146,9 +166,9 @@
146
166
 
147
167
  clf = svm.SVC()
148
168
 
149
- clf.fit(train_data, target_data)
169
+ clf.fit(train_data, label_train)
150
-
170
+
151
- pre = clf.predict(data_test)
171
+ pre = clf.predict()
152
172
 
153
173
 
154
174
 
@@ -164,11 +184,13 @@
164
184
 
165
185
  loo = LeaveOneOut()
166
186
 
187
+
188
+
167
189
  for train, test in loo.split(mnist.data):
168
190
 
169
191
  train_data = mnist.data[train]
170
192
 
171
- target_data = mnist.target[train]
193
+ label_train = mnist.target[train]
172
194
 
173
195
 
174
196
 
@@ -180,9 +202,9 @@
180
202
 
181
203
  model = KNeighborsClassifier(n_neighbors=k)
182
204
 
183
- model.fit(train_data, target_data)
205
+ model.fit(train_data, label_train)
184
-
185
-
206
+
207
+
186
208
 
187
209
  # 一つだけ取り除いたテストデータを識別
188
210
 

4

書式の改善

2017/11/28 08:50

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -124,7 +124,7 @@
124
124
 
125
125
  ```
126
126
 
127
- ###mnistデータで最適なkを探すknnのプログラム現段階
127
+ ###mnistデータで最適なkを探すknnのプログラム　改善版
128
128
 
129
129
  ```python
130
130
 
@@ -134,7 +134,7 @@
134
134
 
135
135
  test_size = 100
136
136
 
137
- data_train, data_test, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
137
+ train_data, test_data_, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
138
138
 
139
139
 
140
140
 
@@ -146,7 +146,7 @@
146
146
 
147
147
  clf = svm.SVC()
148
148
 
149
- clf.fit(data_train, label_train)
149
+ clf.fit(train_data, target_data)
150
150
 
151
151
  pre = clf.predict(data_test)
152
152
 
@@ -166,9 +166,9 @@
166
166
 
167
167
  for train, test in loo.split(mnist.data):
168
168
 
169
- data_train = mnist.data[train]
169
+ train_data = mnist.data[train]
170
-
170
+
171
- data_test = mnist.target[train]
171
+ target_data = mnist.target[train]
172
172
 
173
173
 
174
174
 

3

誤字

2017/11/28 08:26

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -122,12 +122,14 @@
122
122
 
123
123
  main()
124
124
 
125
-
125
+ ```
126
126
 
127
127
  ###mnistデータで最適なkを探すknnのプログラム現段階
128
128
 
129
129
  ```python
130
130
 
131
+
132
+
131
133
  train_size = 500
132
134
 
133
135
  test_size = 100

2

書式の改善

2017/11/28 07:42

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -14,71 +14,13 @@
14
14
 
15
15
  ・容量が大きくて実行できない。
16
16
 
17
- ・ValueError Traceback (most recent call last)
18
-
19
- <ipython-input-10-c9ec06272b57> in <module>()
20
-
21
- 3
22
-
23
- 4 clf = svm.SVC()
24
-
25
- ----> 5 clf.fit(data_train, label_train)
26
-
27
- 6 pre = clf.predict(data_test)
28
-
29
- 7
30
-
31
-
32
-
33
- ~\Anaconda3\lib\site-packages\sklearn\svm\base.py in fit(self, X, y, sample_weight)
34
-
35
- 147 self._sparse = sparse and not callable(self.kernel)
36
-
37
- 148
38
-
39
- --> 149 X, y = check_X_y(X, y, dtype=np.float64, order='C', accept_sparse='csr')
40
-
41
- 150 y = self._validate_targets(y)
42
-
43
- 151
44
-
45
- ~\Anaconda3\lib\site-packages\sklearn\utils\validation.py in check_X_y(X, y, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, multi_output, ensure_min_samples, ensure_min_features, y_numeric, warn_on_dtype, estimator)
46
-
47
- 581 y = y.astype(np.float64)
48
-
49
- 582
50
-
51
- --> 583 check_consistent_length(X, y)
52
-
53
- 584
54
-
55
- 585 return X, y
56
-
57
-
58
-
59
- ~\Anaconda3\lib\site-packages\sklearn\utils\validation.py in check_consistent_length(*arrays)
60
-
61
- 202 if len(uniques) > 1:
62
-
63
- 203 raise ValueError("Found input variables with inconsistent numbers of"
64
-
65
- --> 204 " samples: %r" % [int(l) for l in lengths])
66
-
67
- 205
68
-
69
- 206
70
-
71
-
72
-
73
- ValueError: Found input variables with inconsistent numbers of samples: [69999, 500]
74
-
75
17
 
76
18
 
77
19
 
78
20
 
79
21
  ###Digitsデータで最適なkを探すknnのプログラム
80
22
 
81
-
23
+ ```python
82
24
 
83
25
  from matplotlib import pyplot as plt
84
26
 

1

現段階とエラー

2017/11/28 07:39

投稿

退会済みユーザー
test CHANGED
File without changes
test CHANGED
@@ -12,7 +12,65 @@
12
12
 
13
13
 
14
14
 
15
- 容量が大きくて実行できない。
15
+ ・容量が大きくて実行できない。
16
+
17
+ ・ValueError Traceback (most recent call last)
18
+
19
+ <ipython-input-10-c9ec06272b57> in <module>()
20
+
21
+ 3
22
+
23
+ 4 clf = svm.SVC()
24
+
25
+ ----> 5 clf.fit(data_train, label_train)
26
+
27
+ 6 pre = clf.predict(data_test)
28
+
29
+ 7
30
+
31
+
32
+
33
+ ~\Anaconda3\lib\site-packages\sklearn\svm\base.py in fit(self, X, y, sample_weight)
34
+
35
+ 147 self._sparse = sparse and not callable(self.kernel)
36
+
37
+ 148
38
+
39
+ --> 149 X, y = check_X_y(X, y, dtype=np.float64, order='C', accept_sparse='csr')
40
+
41
+ 150 y = self._validate_targets(y)
42
+
43
+ 151
44
+
45
+ ~\Anaconda3\lib\site-packages\sklearn\utils\validation.py in check_X_y(X, y, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, multi_output, ensure_min_samples, ensure_min_features, y_numeric, warn_on_dtype, estimator)
46
+
47
+ 581 y = y.astype(np.float64)
48
+
49
+ 582
50
+
51
+ --> 583 check_consistent_length(X, y)
52
+
53
+ 584
54
+
55
+ 585 return X, y
56
+
57
+
58
+
59
+ ~\Anaconda3\lib\site-packages\sklearn\utils\validation.py in check_consistent_length(*arrays)
60
+
61
+ 202 if len(uniques) > 1:
62
+
63
+ 203 raise ValueError("Found input variables with inconsistent numbers of"
64
+
65
+ --> 204 " samples: %r" % [int(l) for l in lengths])
66
+
67
+ 205
68
+
69
+ 206
70
+
71
+
72
+
73
+ ValueError: Found input variables with inconsistent numbers of samples: [69999, 500]
16
74
 
17
75
 
18
76
 
@@ -124,6 +182,108 @@
124
182
 
125
183
 
126
184
 
185
+ ###mnistデータで最適なkを探すknnのプログラム現段階
186
+
187
+ ```python
188
+
189
+ train_size = 500
190
+
191
+ test_size = 100
192
+
193
+ data_train, data_test, label_train, label_test = model_selection.train_test_split(mnist_data, mnist_label, test_size=test_size, train_size=train_size)
194
+
195
+
196
+
197
+ K = 10
198
+
199
+ ks = range(1, K + 1)
200
+
201
+
202
+
203
+ clf = svm.SVC()
204
+
205
+ clf.fit(data_train, label_train)
206
+
207
+ pre = clf.predict(data_test)
208
+
209
+
210
+
211
+ # 使う近傍数ごとに正解率＆各経過時間を計算
212
+
213
+ accuracy_scores = []
214
+
215
+ start = time.time()
216
+
217
+ for k in ks:
218
+
219
+ predicted_labels = []
220
+
221
+ loo = LeaveOneOut()
222
+
223
+ for train, test in loo.split(mnist.data):
224
+
225
+ data_train = mnist.data[train]
226
+
227
+ data_test = mnist.target[train]
228
+
229
+
230
+
231
+ elapsed_time = time.time() - start
232
+
233
+
234
+
235
+ # モデルを学習させる
236
+
237
+ model = KNeighborsClassifier(n_neighbors=k)
238
+
239
+ model.fit(train_data, target_data)
240
+
241
+
242
+
243
+ # 一つだけ取り除いたテストデータを識別
244
+
245
+ predicted_label = model.predict(features[test])
246
+
247
+ predicted_labels.append(predicted_label)
248
+
249
+
250
+
251
+ # 正解率を計算
252
+
253
+ ac_score = metrics.accuracy_score(label_test, pre)
254
+
255
+ print('k={0}: {1}'.format(k, ac_score))
256
+
257
+
258
+
259
+ accuracy_scores.append(score)
260
+
261
+
262
+
263
+ # 各経過時間を表示
264
+
265
+ print("経過時間:{0}".format(elapsed_time))
266
+
267
+
268
+
269
+ # 使う近傍数ごとの正解率を折れ線グラフ
270
+
271
+ X = list(ks)
272
+
273
+ plt.plot(X, ac_score)
274
+
275
+
276
+
277
+ plt.xlabel('k')
278
+
279
+ plt.ylabel('accuracy rate')
280
+
281
+ plt.show()
282
+
283
+ ```
284
+
285
+
286
+
127
287
 
128
288
 
129
289
  ###試したこと