使用数据生成器时，在混淆矩阵和准确性方面存在分歧

Question

我正在研究一个模型基于以下代码

epoch=100
model_history = model.fit(train_generator, 
epochs=epoch,
validation_data=test_generator,
callbacks=[model_es, model_rlr, model_mcp])

在模型训练后，当我使用以下代码评估模型时，我得到了98.3%的准确率

model.evaluate(test_generator)

41/41【=========================】-3s 68ms/步-损耗：0.0396-精度：0.9893[0.039571091532707214，0.9893211126327515]

为了分析结果，我试图使用以下代码获得test_generator的混淆矩阵

y_pred = model.predict(test_generator)
y_pred = np.argmax(y_pred, axis=1)
print(confusion_matrix(test_generator.classes, y_pred))

但是输出是

[[ 68  66  93  73]
[ 64  65  93  84]
[ 91 102 126  86]
[ 69  75  96  60]]

这与模型评估高度不一致

有人能帮我获得模型的实际混淆矩阵吗

模型精度的绘图历史

完整代码：https://colab.research.google.com/drive/1wpoPjnSoCqVaA--N04dcUG6A5NEVcufk?usp=sharing

Answer 1

根据您的代码，更改：

test_generator=train_datagen.flow_from_directory(
locat_testing,
class_mode='binary',
color_mode='grayscale',
batch_size=32,
target_size=(img_size,img_size)
)

包括shuffle参数：

test_generator=train_datagen.flow_from_directory(
locat_testing,
class_mode='binary',
color_mode='grayscale',
batch_size=32,
target_size=(img_size,img_size),
shuffle=False
)

你的困惑矩阵看起来会更准确，而不是看起来像随机猜测的。

Answer 2

以下是预测准确性、混淆矩阵和分类报告的代码

def predictor(test_gen):    
y_pred= []
error_list=[]
error_pred_list = []
y_true=test_gen.labels
classes=list(test_gen.class_indices.keys())
class_count=len(classes)
errors=0
preds=model.predict(test_gen, verbose=1)
tests=len(preds)    
for i, p in enumerate(preds):        
pred_index=np.argmax(p)         
true_index=test_gen.labels[i]  # labels are integer values        
if pred_index != true_index: # a misclassification has occurred                                           
errors=errors + 1
file=test_gen.filenames[i]
error_list.append(file)
error_class=classes[pred_index]
error_pred_list.append(error_class)
y_pred.append(pred_index)

acc=( 1-errors/tests) * 100
msg=f'there were {errors} errors in {tests} tests for an accuracy of {acc:6.2f}'
print(msg)
ypred=np.array(y_pred)
ytrue=np.array(y_true)
f1score=f1_score(ytrue, ypred, average='weighted')* 100
if class_count <=30:
cm = confusion_matrix(ytrue, ypred )
# plot the confusion matrix
plt.figure(figsize=(12, 8))
sns.heatmap(cm, annot=True, vmin=0, fmt='g', cmap='Blues', cbar=False)       
plt.xticks(np.arange(class_count)+.5, classes, rotation=90)
plt.yticks(np.arange(class_count)+.5, classes, rotation=0)
plt.xlabel("Predicted")
plt.ylabel("Actual")
plt.title("Confusion Matrix")
plt.show()
clr = classification_report(y_true, y_pred, target_names=classes, digits= 4) # create classification report
print("Classification Report:n----------------------n", clr)
return errors, tests, error_list, error_pred_list, f1score
errors, tests, error_list, error_pred_list, f1score =predictor(test_gen)
# print out list of test files misclassified if less than 50 errors
if len(error_list) > 0 and len(error_list)<50:
print ('Below is a list of test files that were miss classified n')
print ('{0:^30s}{1:^30s}'.format('Test File', ' Predicted as'))
sorted_list=sorted(error_list)
for i in range(len(sorted_list)):
fpath=sorted_list[i]        
split=fpath.split('\')        
f=split[4]+ '-' + split[5]
print(f'{f:^30s}{error_pred_list[i]:^30s}')

Answer 3

# print out list of test files misclassified if less than 50 errors
if len(error_list) > 0 and len(error_list)<50:
print ('Below is a list of test files that were miss classified n')
print ('{0:^30s}{1:^30s}'.format('Test File', ' Predicted as'))
sorted_list=sorted(error_list)
for i in range(len(sorted_list)):
fpath=sorted_list[i]        
split=fpath.split('\')        
f=split[2]+ '-' + split[2]
print(f'{f:^30s}{error_pred_list[i]:^30s}')