featureShardConfigurations when scoring a mixed effects model with random intercept only

Nobody has claimed this yet.

In my first attempt using Photon-ML, I have trained a mixed effects model that includes an intercept-only random effect. I then try to use the scoring driver, using the tutorial as a template.

I'm having trouble specifying the right feature shard for the random intercept. During training, I assign no feature bags to the feature shard (by omitting it in the cli command). In the scoring part, what's the right thing to do?

Thanks in advance!

val globalFeatureShardId = "sGlobal"
val globalFeatureShard = Set("Bu", "Bg", "Bc")

val perUserFeatureShardId = "sUser"
val perUserFeatureShard = Set("uuid")
// only looking for random intercept and therefore the shard is empty, what do I enter here?
// should i put the userid (uuid) in a bag by itself and use it here?

val fixedShardConfigs = Map(globalFeatureShardId -> FeatureShardConfiguration(globalFeatureShard, true))
val mixedShardConfigs = Map(perUserFeatureShardId -> FeatureShardConfiguration(perUserFeatureShard, true)) ++ fixedShardConfigs

GameScoringDriver.set(GameScoringDriver.featureShardConfigurations, mixedShardConfigs)

Start with the tutorial and the GameScoringDriver.featureShardConfigurations entry point, then trace how FeatureShardConfiguration represents an empty random-intercept shard and compare it with the training CLI behavior described here. Done means confirming the correct scoring configuration for an intercept-only random effect, including whether the user ID belongs in a feature bag.

Written by the indexing model from the issue text.