Question

A retail company trains a churn prediction model in Vertex AI using features computed by a nightly BigQuery batch job (e.g., 30-day rolling purchase totals). In production, the online prediction service recomputes those same features with ad hoc SQL written by the application team, and the model's live accuracy is far lower than during offline evaluation. What is the MOST likely root cause and the best remedy?

Accepted Answer

Training-serving skew from divergent feature logic; centralize feature definitions in Vertex AI Feature Store so training and online serving read identical feature values.

Answer

The model is overfitting; increase L2 regularization and reduce the number of features until offline and online accuracy match.

Answer

The batch job is too slow; move feature computation to a streaming Dataflow pipeline to lower latency and restore accuracy.

Answer

The BigQuery training table is stale; enable a materialized view so training always uses the freshest data.