Happy 8th birthday to the Maveryx Community! Take a walk down memory lane in our birthday blog, and don't miss out on the awesome birthday present that all Maveryx Community members get to take advantage of!
I was looking to find if there are any examples or recommendations to get prediction explanations around ML models in Alteryx. What I'm looking for is row by row explanation of which variables contributed most to the prediction made for that row. For instance, the below example uses SHAP, it has the top features and if they contributed positively or negatively to the class prediction:
I'm not aware of any out of the box tools within Alteryx. We use Python and SHAP, or R and the xgboost explainer package (depending on who developed the model). We actually do this outside of Alteryx and within the Python or R script itself (although we do use Alteryx to bring it all together).
We just started exploring SHAP so don't have the details worked out. We have two 3 deployed models that use explainers. 2 of them use the xgboostExplainer package in R, and one uses Lime in R.
We do have a few models built right in Alteryx using their predictive tools (the "brown" R-based ones, not the new blue ones in Intelligence Suite).
All our other models were built in R studio or Python directly. But the process flow once the scoring process is developed is all in Alteryx. It's just that instead of code in the R or Python tool, it's just a call to the R or Python script (.R or .py).