Demographic Predictor Skill

Automatically predict demographics when analyzing person data.

When to Use

This skill is automatically invoked when:

• User provides person data (names, images, profiles)
• User requests demographic analysis
• Batch processing of person records is needed
• Data enrichment for customer profiles is required

Capabilities

Core Predictions

• Age Estimation: Predict age or age range
• Gender Classification: Identify gender
• Ethnicity Analysis: Classify ethnic background
• Name Analysis: Extract demographic signals from names

Advanced Features

• Confidence scoring for all predictions
• Batch processing support
• Multi-modal analysis (name + image + context)
• Historical trend analysis

Tools Available

Use these MCP tools for predictions:

• predict_age: Age estimation from name/image
• predict_gender: Gender classification
• predict_ethnicity: Ethnicity analysis
• analyze_name: Name-based demographic signals
• batch_predict: Process multiple records

Usage Pattern

// Single prediction
const result = await tools.predict_age({
  name: "John Smith",
  image_url: "https://...",
  context: "LinkedIn profile"
});

// Batch prediction
const results = await tools.batch_predict({
  records: [
    { id: 1, name: "Jane Doe" },
    { id: 2, name: "Carlos Rodriguez" }
  ],
  predictions: ["age", "gender", "ethnicity"]
});

Output Format

Always return structured predictions:

{
  "predictions": {
    "age": {
      "value": "25-34",
      "confidence": 0.85,
      "method": "name_analysis"
    },
    "gender": {
      "value": "female",
      "confidence": 0.92,
      "method": "name_analysis"
    },
    "ethnicity": {
      "value": "hispanic",
      "confidence": 0.78,
      "method": "name_analysis"
    }
  },
  "metadata": {
    "timestamp": "2025-11-08T...",
    "model_version": "0.1.0",
    "processing_time_ms": 150
  }
}

Best Practices

1. Always provide confidence scores - Users need to understand prediction reliability
2. Document data sources - Track what information was used
3. Handle missing data gracefully - Work with incomplete information
4. Respect privacy - Never expose PII unnecessarily
5. Flag low confidence - Warn when predictions are uncertain

Error Handling

• Return null values for unpredictable attributes
• Set confidence to 0 when data is insufficient
• Log errors without exposing sensitive data
• Suggest data enrichment when needed

Performance

• Single predictions: < 200ms target
• Batch predictions: < 2s for 100 records
• Cache common patterns for speed
• Use async processing for large batches