Predictions for the model

Hello,

I was exploring the predict method for the model. So the predict gives two values, first is the array of 1s and 0s and second is the values of floats.

I have a few questions.

1. How do you calculate the 0s and 1s array, is it based on some probability or what?
2. What are the values in the float array? Are they activation values? Also, how can I convert them to probabilities? Doing a softmax over them won’t make sense because the model is trained in a 1vsall way.

Thanks

[kinoute]

Hi,

To convert the raw_outputs, you can use softmax from the scipy library:

from scipy.special import softmax

probabilities = softmax(raw_outputs, axis=1)

It will convert the float numbers to predictions like [0.12374 0.87612]

[ThilinaRajapakse]

Yes.

You’d be right to assume that I know what the 1 vs all approach is. What I pointed out was that if the labels are independent of each other, you cannot select a top k based on the scores given by the model for each label as they are independent of each other.