Partitioning and Commas in Partition Names

[Rand3rino]

Hello everyone,

I am building a partitioned model by customer name and I'm running into a challenge when scoring new data.

For example, I have a customer: Business Entity, LLC. that when I build the models, there is a model for exactly "Business Entity, LLC". However when scoring, the jobs calls two models "Business Entity" and "LLC" because of the comma and therefore failing because I don't have models with those partition names.

How do I get around this? Are there escape characters I can use?

[Turribeach · Accepted solution]

You could remove the commas from the Customer Name in a prepare recipe.

[Rand3rino]

Expanding on the solution here, discrete partition names cannot have special characters or punctuation.

Partition identifiers — Dataiku DSS 12 documentation