Overview

Overview#

The Data Product Recommender analyzes query logs to identify high-value tables for data product creation.

Scoring Methodology#

Individual Table Scoring (0-100)#

  • 37.5% Query Count - Volume of usage

  • 37.5% User Diversity - Breadth of usage across teams

  • 15% Recency - Recent activity

  • 10% Consistency - Regular usage patterns

Table Group Scoring (0-100)#

  • 30% Cohesion - How tightly tables are connected

  • 20% Usage - Relative usage compared to other groups

  • 15% User Reach - Percentage of users querying the group

  • 20% Recency - Recent activity across tables

  • 10% Consistency - Regular usage patterns

  • 5% Size - Number of tables in the group

Star Rating Scale#

  • ⭐⭐⭐⭐⭐ Excellent (80-100): Implement immediately

  • ⭐⭐⭐⭐ Good (60-79): Medium priority

  • ⭐⭐⭐ Fair (40-59): Consider splitting or implement later

  • ⭐⭐ Weak (20-39): Reconsider grouping

  • ⭐ Poor (0-19): Do not implement

Platform Support#

  • ✅ Snowflake - Export from SNOWFLAKE.ACCOUNT_USAGE.QUERY_HISTORY

  • ✅ Databricks - Export from system.query.history

  • ✅ BigQuery - Export from INFORMATION_SCHEMA.JOBS_BY_PROJECT

  • ✅ watsonx.data - Export from system.runtime.queries

See Also#