Custom Metrics for Keras and TensorFlow

Hi, thanks for sharing. I'm new with DL, i would ask how to calculate the y_true?

Author

the y_true is the ground-truth of your dataset.

the y_true is the ground-truth of your dataset.

thanks for your reply, i faced an error and solved by changing the datatype of the label dataset. My question is after called the precision when i tried to "print" the value it showed nothing. I working in an ensemble learning model . sorry for my trivial question.
regards

Hi Arnal,
I realized that my balanced accuracy values calculated within tensorflow are not the same ones calculated by the regular confusion matrix from the caret package (it's in R, but the code is very similar). Have you had any experience with that in python? I tried both codes for balanced accuracy:

Tensorflow

balanced_acc <- custom_metric("balanced_acc",function(y_true,y_pred){
  y_pred_pos = k_round(k_clip(y_pred, 0, 1))
  y_pred_neg = 1 - y_pred_pos
  
  y_pos = k_round(k_clip(y_true, 0, 1))
  y_neg = 1 - y_pos
  
  tp = k_sum(y_pos * y_pred_pos)
  tn = k_sum(y_neg * y_pred_neg)
  
  fp = k_sum(y_neg * y_pred_pos)
  fn = k_sum(y_pos * y_pred_neg)
  
  sensi = (tp/(tp + fn + k_epsilon()))
  specifi = (tn/(tn + fp + k_epsilon()))
  
  return((sensi + specifi )/ 2 )
  
})


##or 

balanced_acc <- custom_metric("balanced_acc",function(y_true,y_pred){

  tp = k_sum(k_round(k_clip(y_true * y_pred, 0, 1)))
  tn =  k_sum(k_round(k_clip((1 - y_true) * (1 - y_pred), 0, 1)))
  
  fp = k_sum(k_round(k_clip((1 - y_true) * y_pred, 0, 1)))
  fn = k_sum(k_round(k_clip(y_true * (1 - y_pred), 0, 1)))
  
  sensi = (tp/(tp + fn + k_epsilon()))
  specifi = (tn/(tn + fp + k_epsilon()))
  
  return((sensi + specifi )/ 2 )
  
})

caret

  conf.matrix <-  caret::confusionMatrix(
    factor(pred_model, levels = 0:1),
    factor(y_test, levels = 0:1),
    positive = "1"
  )
  
  BalancedAccuracy <- conf.matrix$byClass[11]

Author

Hi, @DohaNaga
I would check the shape of each matrix and if they are disposed in the same way (i.e., if rows are samples and cols are features).