Model Predictions Returning Zero Values – Need Help with Training Input Setup
Hi everyone, I’m encountering an issue with my predictive model in Qlik. I followed the standard workflow created an AutoML experiment, deployed a model version, ran predictions, and then created an analytics app to view the results.
However, when I check the predictions for my target column (which is a numerical value), all the values are showing up as zero.
Here’s a brief about my dataset: • Total of 4 columns: • 1 numerical column as the target. • 2 categorical columns. • 1 column with dates.
I’m unsure whether I’ve set up the training dataset and the prediction input dataset correctly.
My key doubts: • What should be included in the training set vs. the prediction input set? • Could this issue (zero predictions) be due to a mismatch in feature formatting or data leakage? • Has anyone faced a similar problem where predictions return only zeros after model deployment?
Would really appreciate any guidance or tips on how to troubleshoot or correctly configure the pipeline.Thanks in advance!
I think we are going to need more details to be able help, but are sure the column is returning zero's? or it maybe just a formatting issue on load? trying wrapping in the num function.
I feel if it was a modelling issue you would have seen this in the results when building the model.
