質問編集履歴
2
文法修正
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -140,4 +140,22 @@
|
|
|
140
140
|
|
|
141
141
|
if __name__ == "__main__":
|
|
142
142
|
main()
|
|
143
|
+
```
|
|
144
|
+
|
|
145
|
+
```ここに言語を入力
|
|
146
|
+
Ypred [1 1 1 ... 6 6 6]
|
|
147
|
+
|
|
148
|
+
Ytest
|
|
149
|
+
384 1
|
|
150
|
+
385 1
|
|
151
|
+
386 1
|
|
152
|
+
387 1
|
|
153
|
+
388 1
|
|
154
|
+
..
|
|
155
|
+
7675 8
|
|
156
|
+
7676 8
|
|
157
|
+
7677 8
|
|
158
|
+
7678 8
|
|
159
|
+
7679 8
|
|
160
|
+
Name: category, Length: 1536, dtype: in
|
|
143
161
|
```
|
1
改善
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -23,4 +23,121 @@
|
|
|
23
23
|
print(precision_score(Y_test, Y_pred,average=’micro’))
|
|
24
24
|
^
|
|
25
25
|
SyntaxError: invalid character in identifier
|
|
26
|
+
```
|
|
27
|
+
|
|
28
|
+
コード本文です。
|
|
29
|
+
```ここに言語を入力
|
|
30
|
+
# -*- coding: utf-8 -*-
|
|
31
|
+
|
|
32
|
+
import scipy as sp
|
|
33
|
+
import pandas as pd
|
|
34
|
+
from pandas import Series, DataFrame
|
|
35
|
+
import matplotlib.pyplot as plt
|
|
36
|
+
from sklearn.preprocessing import StandardScaler
|
|
37
|
+
from sklearn.ensemble import RandomForestClassifier #RandomForest
|
|
38
|
+
from sklearn.svm import SVC # SVM用
|
|
39
|
+
import lightgbm as lgb
|
|
40
|
+
from sklearn.metrics import confusion_matrix #混同行列
|
|
41
|
+
from sklearn.metrics import accuracy_score, precision_score #適合率
|
|
42
|
+
from sklearn.metrics import recall_score, f1_score #再現率,F1スコア
|
|
43
|
+
from sklearn.metrics import make_scorer
|
|
44
|
+
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
|
|
49
|
+
|
|
50
|
+
|
|
51
|
+
def SVM(X_train_std, X_test_std, Y_train, Y_test):
|
|
52
|
+
model_SVM = SVC(random_state=0, kernel = "rbf", C = 1000 , gamma = 0.01 )
|
|
53
|
+
#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
|
|
54
|
+
model_SVM.fit(X_train_std,Y_train)
|
|
55
|
+
|
|
56
|
+
#予測したクラスラベル
|
|
57
|
+
Y_pred = model_SVM.predict(X_test_std)
|
|
58
|
+
|
|
59
|
+
# .scoreで正解率を算出。
|
|
60
|
+
print("\nSVM")
|
|
61
|
+
print("train score:",model_SVM.score(X_train_std,Y_train))
|
|
62
|
+
print("test score:",model_SVM.score(X_test_std,Y_test))
|
|
63
|
+
|
|
64
|
+
print("accuracy score:",accuracy_score(Y_test, Y_pred,average='micro'))
|
|
65
|
+
print("precision score:",precision_score(Y_test, Y_pred,average='micro'))
|
|
66
|
+
print("recall score:",recall_score(Y_test, Y_pred,average='micro'))
|
|
67
|
+
print("f1 score:",f1_score(Y_test, Y_pred,average='micro'))
|
|
68
|
+
|
|
69
|
+
|
|
70
|
+
|
|
71
|
+
def GBM(X_train, X_test, Y_train, Y_test):
|
|
72
|
+
model_GBM = lgb.LGBMClassifier(boosting_type='gbdt', num_leaves=58,
|
|
73
|
+
max_depth=14, learning_rate=0.1, n_estimators=940,
|
|
74
|
+
min_child_samples=40, importance_type="split", random_state=0)
|
|
75
|
+
|
|
76
|
+
#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
|
|
77
|
+
model_GBM.fit(X_train, Y_train)
|
|
78
|
+
|
|
79
|
+
#予測したクラスラベル
|
|
80
|
+
Y_pred = model_GBM.predict(X_test)
|
|
81
|
+
|
|
82
|
+
# .scoreで正解率を算出。
|
|
83
|
+
print("\nGBM")
|
|
84
|
+
print("train score:",model_GBM.score(X_train,Y_train))
|
|
85
|
+
print("test score:",model_GBM.score(X_test,Y_test))
|
|
86
|
+
|
|
87
|
+
print("accuracy score:",accuracy_score(Y_test, Y_pred,average='micro'))
|
|
88
|
+
print("precision score:",precision_score(Y_test, Y_pred,average='micro'))
|
|
89
|
+
print("recall score:",recall_score(Y_test, Y_pred,average='micro'))
|
|
90
|
+
print("f1 score:",f1_score(Y_test, Y_pred,average='micro'))
|
|
91
|
+
|
|
92
|
+
|
|
93
|
+
|
|
94
|
+
|
|
95
|
+
|
|
96
|
+
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
def main():
|
|
100
|
+
"""
|
|
101
|
+
TODO case_nameに任意の名前を指定
|
|
102
|
+
フォルダを統一するために以降のscriptも名前を統一する
|
|
103
|
+
"""
|
|
104
|
+
case_name = "case3"
|
|
105
|
+
# ---------------------------------------
|
|
106
|
+
|
|
107
|
+
case_dir = "./casestudy/{}/".format(case_name)
|
|
108
|
+
input_csv_name = "2_extracted_features_original.csv"
|
|
109
|
+
input_csv_path = case_dir + input_csv_name
|
|
110
|
+
input_df = pd.read_csv(input_csv_path, encoding="utf-8-sig")
|
|
111
|
+
|
|
112
|
+
|
|
113
|
+
#すべてのデータを対象に分類を行う場合
|
|
114
|
+
#-----------------------------------------------------------------------
|
|
115
|
+
task = "all"
|
|
116
|
+
train_df = input_df[input_df["train_test_flag"] == 0]
|
|
117
|
+
test_df = input_df[input_df["train_test_flag"] == 1]
|
|
118
|
+
# print(train_df)
|
|
119
|
+
# print(test_df)
|
|
120
|
+
|
|
121
|
+
X_train = train_df.loc[:, "contrast":"inverse_difference_m_norm"]
|
|
122
|
+
X_test = test_df.loc[:, "contrast":"inverse_difference_m_norm"]
|
|
123
|
+
# print(X_train)
|
|
124
|
+
# print(X_test)
|
|
125
|
+
Y_train = train_df["category"]
|
|
126
|
+
Y_test = test_df["category"]
|
|
127
|
+
# print(Y_train)
|
|
128
|
+
# print(Y_test)
|
|
129
|
+
|
|
130
|
+
sc = StandardScaler()
|
|
131
|
+
sc.fit(X_train)
|
|
132
|
+
X_train_std = sc.transform(X_train)
|
|
133
|
+
X_test_std = sc.transform(X_test)
|
|
134
|
+
# print(X_train_std)
|
|
135
|
+
# print(X_test_std)
|
|
136
|
+
# #------------------------------------------------------------------------
|
|
137
|
+
SVM(X_train_std, X_test_std, Y_train, Y_test)
|
|
138
|
+
GBM(X_train, X_test, Y_train, Y_test)
|
|
139
|
+
|
|
140
|
+
|
|
141
|
+
if __name__ == "__main__":
|
|
142
|
+
main()
|
|
26
143
|
```
|