# How to calculate entropy of each class to measure model uncertainty

I am trying to calculate Entropy to measure model’s uncertainty using MC Dropout for image classification task ,I have calculated the Entropy for each sample using the samples mean output_mean as shown in the code below

``````for images,labels in testloader:
images = images.to(device)
labels = labels.to(device)

output_list = []
for i in range(T):
output_list.append(torch.unsqueeze(model(images), 0))

#calculating samples mean
output_mean = torch.cat(output_list, 0).mean(0) #shape (n_samples, n_classes)
output_mean = np.asarray(output_mean.cpu())

epsilon = sys.float_info.min
# Calculating entropy across multiple MCD forward passes
entropy = -np.sum(output_mean*np.log(output_mean + epsilon), axis=-1) #shape (n_samples, n_classes)
``````

After calculating Entropy of each sample, I am trying to calculate Entropy for each class to get the model uncertainty about each one of them. Can anyone help me to get the right formula to calculate entropy