Hi everyone,
I’m currently working with both Artificial Neural Network (ANN) and XGBoost models in JMP Pro 18, and I’ve encountered a recurring issue related to saving predicted values.
After training the models, the internal performance metrics (e.g., R² > 0.98) and the actual vs. predicted plots indicate excellent model performance and a strong 1:1 alignment. However, when I use the “Save Formula” option from the model's red triangle to generate predicted values in the data table, the results do not match those shown during model evaluation. In fact, most of the saved predictions are significantly off.
Thank you in advance for your guidance.
Ali
