starlake-ai/metrics

Metrics Skill

Computes statistical metrics on a table's data. Metrics are based on the metric type defined on each attribute: continuous for numeric columns (min, max, mean, stddev, etc.) and discrete for categorical columns (distinct count, frequency distribution).

Usage

starlake metrics [options]

Options

• --domain <value>: Domain name (required)
• --schema <value>: Table/schema name (required)
• --authInfo k1=v1,k2=v2: Auth info for the connection (e.g. gcpProjectId=my-project)
• --reportFormat <value>: Report output format: console, json, or html

Configuration Context

Attribute Metric Types

Metrics are configured per attribute in table YAML files:

# In table.sl.yml
table:
  attributes:
    - name: "total_amount"
      type: "decimal"
      metric: "continuous"   # min, max, mean, median, variance, stddev, percentiles
    - name: "status"
      type: "string"
      metric: "discrete"     # count distinct, category frequency
    - name: "order_id"
      type: "long"
      # No metric - not computed

Metric Types

| Metric Type | Computed Values | |---|---| | continuous | min, max, sum, mean, median, variance, stddev, skewness, kurtosis, 25th/75th percentiles, missing values, row count | | discrete | count distinct, category frequency, top categories, row count |

Application-Level Configuration

# metadata/application.sl.yml
application:
  metrics:
    active: true
    discreteMaxCardinality: 10    # Max distinct values for discrete metrics
    path: "{{SL_ROOT}}/metrics"   # Custom metrics output path

Metrics Storage

Metrics are stored in the SL_METRICS audit table for historical tracking and analysis. This enables:

• Trend analysis of data quality over time
• Alerting on metric threshold violations
• Data profiling dashboards

Examples

Compute Metrics for a Table

starlake metrics --domain starbake --schema orders

Compute with Auth Info

starlake metrics --domain starbake --schema orders --authInfo gcpProjectId=my-gcp-project

Compute with JSON Report

starlake metrics --domain starbake --schema products --reportFormat json

Related Skills

• summarize - Display table summary statistics
• freshness - Check data freshness
• config - Configuration reference (metrics settings)