skills/data-analysis/references/analysis-patterns.md

# Analysis Patterns

## Comparison

- Calculate absolute values AND relative differences (% change)
- Rank items from highest to lowest
- Highlight the most significant differences
- Use markdown tables for side-by-side presentation

## Trend

- Calculate period-over-period changes
- CAGR for long periods (3+ years)
- Identify seasonality patterns
- Note inflection points (where direction changes)
- Present chronologically

## Distribution

- Calculate mean, median, mode, standard deviation
- Identify outliers beyond 2 standard deviations
- Note skewness (left/right-leaning)
- Present with ranges and percentiles when useful

## Correlation

- Calculate correlation coefficients
- State direction (positive/negative) and strength (weak/moderate/strong)
- Always caveat: correlation ≠ causation
- Present scatter-plot-style observations when multiple data points exist

## Composition

- Calculate each component's share as both absolute value and percentage
- Order from largest to smallest
- Group small components (<5%) into "Other" if there are many
- Present as tables with both raw numbers and percentages

## Presentation Rules

- **Always present data as markdown tables**, never as raw JSON
- JSON output only if user explicitly requests raw data or a code block
- Lead with the key insight or takeaway
- Follow with supporting data and breakdowns
- Bold the most important numbers