cannot build confusion matrix

Question

I want to build a confusion matrix for my CNN model there is the code:

classifier = Sequential()


classifier.add(Conv2D(32, (3, 3), input_shape=(64,64, 3), 
activation='relu'))


classifier.add(MaxPooling2D(pool_size=(2, 2)))  

classifier.add(Flatten())
classifier.add(Dense(units=1, activation='sigmoid'))



classifier.compile(optimizer='adam', loss='binary_crossentropy', metrics= 
['accuracy'])


batch_size = 32

train_datagen = ImageDataGenerator(rescale=1. / 255,
                               shear_range=0.2,
                               zoom_range=0.2,
                               horizontal_flip=True)

test_datagen = ImageDataGenerator(rescale=1. / 255)

training_set = train_datagen.flow_from_directory('x1' ,
                                             target_size=(64,64),
                                             batch_size=64,
                                             class_mode='binary')

test_set = test_datagen.flow_from_directory('x2' ,
                                        target_size=(64,64),
                                        batch_size=64,
                                        class_mode='binary')



ep=50

H=classifier.fit_generator(training_set,
                     steps_per_epoch=1204/batch_size,
                     epochs=ep,
                     validation_data=test_set,
                     validation_steps=408/batch_size,
                                     )

validation_steps=408

confusion matrix:

from sklearn.metrics import confusion_matrix
Y_pred = classifier.predict_generator(test_set,validation_steps//batch_size+1)


y_pred = np.argmax(Y_pred, axis=1)
print('Confusion Matrix')
print(confusion_matrix(test_set.classes, y_pred))

I got this error :

ValueError: Found input variables with inconsistent numbers of samples: [408, 792]

What should I do?

Try model.predict_generator(test_generator,steps = len(test_set)) and check the shapes — Sreeram TP
– Sreeram TP, Commented Mar 13, 2019 at 11:28

Karl · Accepted Answer · 2019-03-13 11:07:03Z

1

test_set and y_pred don't have the same number of observations. Likely the number of steps you are passing to predict_generator is incorrect.

Not sure which version of Keras you are using, but try Y_pred =classifier.predict_generator(test_set). In newer versions leaving out number of steps will return predictions on all images.

answered Mar 13, 2019 at 11:07

Karl

5,84210 gold badges53 silver badges79 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

nima nima Over a year ago

I use this it gives me confusion matrix but the confusion matrix it's not correct because it has two classes it returns [[ 42 0] [366 0]] each of them are the number of items that are in two classes, this means it predicts all of them correct? if it is true why accrucy is not 100 %

Karl Over a year ago

No, that doesn't mean it's 100% correct. It means it is classifying every single observation into the same class. Top-left and bottom-right are correct classifications, so 366 are being classified incorrectly.

nima nima Over a year ago

why all of them classified into one class? is my code incorrect? could you please take a look at my code @karl

Karl Over a year ago

It means your model has learnt it is easier to simply classify everything as one class rather than trying to split them. There could be a hundred reasons for this. Your model might be too simple, your data might not be good enough, your learning rate might be too high, etc...

Ioannis Nasios · Accepted Answer · 2019-03-13 12:02:20Z

1

You have

batch_size = 32

but inside test_set,

batch_size=64

so when you run predict_generator with validation_steps//batch_size+1 steps you are using 32 to create your steps but 64 for your generator.

You must set both to 32 or 64

answered Mar 13, 2019 at 12:02

Ioannis Nasios

8,5474 gold badges41 silver badges59 bronze badges

Collectives™ on Stack Overflow

cannot build confusion matrix

2 Answers 2

4 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

4 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related