python實(shí)現(xiàn)二分類和多分類的ROC曲線教程

更新時間：2020年06月15日 14:36:30 作者：Hinton-wu

這篇文章主要介紹了python實(shí)現(xiàn)二分類和多分類的ROC曲線教程，具有很好的參考價值，希望對大家有所幫助。一起跟隨小編過來看看吧

基本概念

precision：預(yù)測為對的當(dāng)中，原本為對的比例（越大越好，1為理想狀態(tài)）

recall：原本為對的當(dāng)中，預(yù)測為對的比例（越大越好，1為理想狀態(tài)）

F-measure：F度量是對準(zhǔn)確率和召回率做一個權(quán)衡（越大越好，1為理想狀態(tài)，此時precision為1，recall為1）

accuracy：預(yù)測對的（包括原本是對預(yù)測為對，原本是錯的預(yù)測為錯兩種情形）占整個的比例（越大越好，1為理想狀態(tài)）

fp rate：原本是錯的預(yù)測為對的比例（越小越好，0為理想狀態(tài)）

tp rate：原本是對的預(yù)測為對的比例（越大越好，1為理想狀態(tài)）

ROC曲線通常在Y軸上具有真陽性率，在X軸上具有假陽性率。這意味著圖的左上角是“理想”點(diǎn) - 誤報(bào)率為零，真正的正率為1。這不太現(xiàn)實(shí)，但它確實(shí)意味著曲線下面積（AUC）通常更好。

二分類問題：ROC曲線

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import time
start_time = time.time()
import matplotlib.pyplot as plt
from sklearn.metrics import roc_curve
from sklearn.metrics import auc
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.metrics import recall_score,accuracy_score
from sklearn.metrics import precision_score,f1_score
from keras.optimizers import Adam,SGD,sgd
from keras.models import load_model

print('讀取數(shù)據(jù)')
X_train = np.load('x_train-rotate_2.npy')
Y_train = np.load('y_train-rotate_2.npy')
print(X_train.shape)
print(Y_train.shape)

print('獲取測試數(shù)據(jù)和驗(yàn)證數(shù)據(jù)')
X_train, X_valid, Y_train, Y_valid = train_test_split(X_train, Y_train, test_size=0.1, random_state=666)

Y_train = np.asarray(Y_train,np.uint8)
Y_valid = np.asarray(Y_valid,np.uint8)
X_valid = np.array(X_valid, np.float32) / 255.

print('獲取模型')
model = load_model('./model/InceptionV3_model.h5')
opt = Adam(lr=1e-4)
model.compile(optimizer=opt, loss='binary_crossentropy')

print("Predicting")
Y_pred = model.predict(X_valid)
Y_pred = [np.argmax(y) for y in Y_pred] # 取出y中元素最大值所對應(yīng)的索引
Y_valid = [np.argmax(y) for y in Y_valid]

# micro：多分類　　
# weighted：不均衡數(shù)量的類來說，計(jì)算二分類metrics的平均
# macro：計(jì)算二分類metrics的均值，為每個類給出相同權(quán)重的分值。
precision = precision_score(Y_valid, Y_pred, average='weighted')
recall = recall_score(Y_valid, Y_pred, average='weighted')
f1_score = f1_score(Y_valid, Y_pred, average='weighted')
accuracy_score = accuracy_score(Y_valid, Y_pred)
print("Precision_score:",precision)
print("Recall_score:",recall)
print("F1_score:",f1_score)
print("Accuracy_score:",accuracy_score)

# 二分類?。遥希们€
# roc_curve:真正率（True Positive Rate , TPR）或靈敏度（sensitivity）
# 橫坐標(biāo)：假正率（False Positive Rate , FPR）
fpr, tpr, thresholds_keras = roc_curve(Y_valid, Y_pred)
auc = auc(fpr, tpr)
print("AUC : ", auc)
plt.figure()
plt.plot([0, 1], [0, 1], 'k--')
plt.plot(fpr, tpr, label='Keras (area = {:.3f})'.format(auc))
plt.xlabel('False positive rate')
plt.ylabel('True positive rate')
plt.title('ROC curve')
plt.legend(loc='best')
plt.savefig("../images/ROC/ROC_2分類.png")
plt.show()

print("--- %s seconds ---" % (time.time() - start_time))

ROC圖如下所示：

多分類問題：ROC曲線

ROC曲線通常用于二分類以研究分類器的輸出。為了將ROC曲線和ROC區(qū)域擴(kuò)展到多類或多標(biāo)簽分類，有必要對輸出進(jìn)行二值化。⑴可以每個標(biāo)簽繪制一條ROC曲線。⑵也可以通過將標(biāo)簽指示符矩陣的每個元素視為二元預(yù)測（微平均）來繪制ROC曲線。⑶另一種用于多類別分類的評估方法是宏觀平均，它對每個標(biāo)簽的分類給予相同的權(quán)重。

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import time
start_time = time.time()
import matplotlib.pyplot as plt
from sklearn.metrics import roc_curve
from sklearn.metrics import auc
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.metrics import recall_score,accuracy_score
from sklearn.metrics import precision_score,f1_score
from keras.optimizers import Adam,SGD,sgd
from keras.models import load_model
from itertools import cycle
from scipy import interp
from sklearn.preprocessing import label_binarize

nb_classes = 5
print('讀取數(shù)據(jù)')
X_train = np.load('x_train-resized_5.npy')
Y_train = np.load('y_train-resized_5.npy')
print(X_train.shape)
print(Y_train.shape)

print('獲取測試數(shù)據(jù)和驗(yàn)證數(shù)據(jù)')
X_train, X_valid, Y_train, Y_valid = train_test_split(X_train, Y_train, test_size=0.1, random_state=666)

Y_train = np.asarray(Y_train,np.uint8)
Y_valid = np.asarray(Y_valid,np.uint8)
X_valid = np.asarray(X_valid, np.float32) / 255.

print('獲取模型')
model = load_model('./model/SE-InceptionV3_model.h5')
opt = Adam(lr=1e-4)
model.compile(optimizer=opt, loss='categorical_crossentropy')

print("Predicting")
Y_pred = model.predict(X_valid)
Y_pred = [np.argmax(y) for y in Y_pred] # 取出y中元素最大值所對應(yīng)的索引
Y_valid = [np.argmax(y) for y in Y_valid]

# Binarize the output
Y_valid = label_binarize(Y_valid, classes=[i for i in range(nb_classes)])
Y_pred = label_binarize(Y_pred, classes=[i for i in range(nb_classes)])

# micro：多分類　　
# weighted：不均衡數(shù)量的類來說，計(jì)算二分類metrics的平均
# macro：計(jì)算二分類metrics的均值，為每個類給出相同權(quán)重的分值。
precision = precision_score(Y_valid, Y_pred, average='micro')
recall = recall_score(Y_valid, Y_pred, average='micro')
f1_score = f1_score(Y_valid, Y_pred, average='micro')
accuracy_score = accuracy_score(Y_valid, Y_pred)
print("Precision_score:",precision)
print("Recall_score:",recall)
print("F1_score:",f1_score)
print("Accuracy_score:",accuracy_score)

# roc_curve:真正率（True Positive Rate , TPR）或靈敏度（sensitivity）
# 橫坐標(biāo)：假正率（False Positive Rate , FPR）

# Compute ROC curve and ROC area for each class
fpr = dict()
tpr = dict()
roc_auc = dict()
for i in range(nb_classes):
 fpr[i], tpr[i], _ = roc_curve(Y_valid[:, i], Y_pred[:, i])
 roc_auc[i] = auc(fpr[i], tpr[i])

# Compute micro-average ROC curve and ROC area
fpr["micro"], tpr["micro"], _ = roc_curve(Y_valid.ravel(), Y_pred.ravel())
roc_auc["micro"] = auc(fpr["micro"], tpr["micro"])

# Compute macro-average ROC curve and ROC area

# First aggregate all false positive rates
all_fpr = np.unique(np.concatenate([fpr[i] for i in range(nb_classes)]))

# Then interpolate all ROC curves at this points
mean_tpr = np.zeros_like(all_fpr)
for i in range(nb_classes):
 mean_tpr += interp(all_fpr, fpr[i], tpr[i])

# Finally average it and compute AUC
mean_tpr /= nb_classes

fpr["macro"] = all_fpr
tpr["macro"] = mean_tpr
roc_auc["macro"] = auc(fpr["macro"], tpr["macro"])

# Plot all ROC curves
lw = 2
plt.figure()
plt.plot(fpr["micro"], tpr["micro"],
  label='micro-average ROC curve (area = {0:0.2f})'
  ''.format(roc_auc["micro"]),
  color='deeppink', linestyle=':', linewidth=4)

plt.plot(fpr["macro"], tpr["macro"],
  label='macro-average ROC curve (area = {0:0.2f})'
  ''.format(roc_auc["macro"]),
  color='navy', linestyle=':', linewidth=4)

colors = cycle(['aqua', 'darkorange', 'cornflowerblue'])
for i, color in zip(range(nb_classes), colors):
 plt.plot(fpr[i], tpr[i], color=color, lw=lw,
  label='ROC curve of class {0} (area = {1:0.2f})'
  ''.format(i, roc_auc[i]))

plt.plot([0, 1], [0, 1], 'k--', lw=lw)
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Some extension of Receiver operating characteristic to multi-class')
plt.legend(loc="lower right")
plt.savefig("../images/ROC/ROC_5分類.png")
plt.show()

print("--- %s seconds ---" % (time.time() - start_time))

ROC圖如下所示：