I've been working with the EBK Regression Prediction tool inside the Geostatistical Wizard, and I find it very powerful. One feature I particularly appreciate is how quickly I can get cross-validation statistics without having to export full output surfaces — this is great for testing covariates and neighborhood settings interactively.
However, once the model looks promising, I need the actual physical layers. So I use GA Layers to Rasters to export the prediction raster — which, by the way, isn't particularly fast.
Now that I’ve run several tests, I’d like to switch to Python to iterate over multiple fields and covariate combinations. But when looking at the documentation for arcpy.ga.EBKRegressionPrediction_ga(), I can’t figure out how to generate additional outputs like:
Standard error
Probability surfaces
Quantile surfaces
From the Wizard, I can generate these via GA Layers to Rasters, but in Python, I don’t see how to request them — nor do I see parameters like output_type or quantile_value.
I’m also a bit confused about the out_ga_layer parameter. It doesn’t seem to control the output type. So what is its purpose exactly in this context? How do I specify what kind of output I want?
One strength of the tool is the ability to generate the Output diagnostic feature class. On the other hand, I can't see how to get the cross validation stats in Python, I assume I can still get it using out_diagnostic_feature_class, and then calculate metrics myself — but there's no built-in summary like in the Wizard.
To sum up: It seems that when using the Wizard + GA Layers to Rasters, I can access all expected outputs (prediction, standard error, probability, quantiles). But if I want to run the same process from Python, I can only get the prediction raster — and I lose both the optional outputs and the summary cross-validation stats.
My question is: Is there any way to programmatically generate all outputs available through the Wizard using EBKRegressionPrediction_ga() in Python?
