Amazon Machine Learning Studio: Adjusted Score threshold is not applied while making real time predictions

Question

I am using Amazon Machine Learning Studio's binary classification algorithm, wherein I have adjusted my score threshold to 0.01 as per my requirement. Thus, all the records for which the prediction score is above 0.01 should predict 1. However, while making the real-time prediction, despite the computed prediction score being greater than 0.01, the prediction label is returned as 0.

Can someone explain it to me why is the adjusted threshold not applied while making the predictions?

Answer 1

For the adjusted threshold to be applied while making the predictions, we need to re-evaluate the model. The predictions made after re-evaluation take the updated score threshold into consideration while computing the prediction label.