aget-analyze-data

Perform analysis on datasets to discover patterns, trends, and anomalies. Generate actionable insights from data.

Instructions

When this skill is invoked:

1. Profile the Dataset

   • Identify data structure (fields, types, record count)
   • Assess data quality (completeness, consistency)
   • Note any limitations or caveats

2. Compute Statistics

   • For numerical fields: mean, median, range, distribution
   • For categorical fields: frequency counts, top values
   • For temporal fields: trends, seasonality

3. Discover Patterns

   • Identify correlations between fields
   • Detect outliers and anomalies
   • Find trends over time (if applicable)

4. Generate Insights

   • Translate findings into actionable statements
   • Prioritize by potential impact
   • Distinguish correlation from causation

Output Format

## Data Analysis: [Dataset/Topic]

### Dataset Profile

| Attribute | Value |
|-----------|-------|
| Records | [N] |
| Fields | [N] |
| Date Range | [Start - End] |
| Quality | [Good/Fair/Poor] |

### Key Statistics

| Field | Type | Summary |
|-------|------|---------|
| [Field 1] | Numeric | Mean: X, Median: Y, Range: [A-B] |
| [Field 2] | Categorical | Top: [Value] (N%), [Value] (N%) |

### Patterns Discovered

1. **[Pattern Name]**: [Description]
   - Evidence: [Data points supporting this]
   - Confidence: [High/Medium/Low]

### Anomalies

- [Anomaly 1]: [Description and potential significance]

### Actionable Insights

1. **[Insight]**: [Recommended action based on finding]

### Limitations

- [Limitation 1]: [How it affects conclusions]

Constraints

• C1: NEVER present correlation as causation — statistical rigor required
• C2: NEVER ignore outliers without explicit justification — anomalies may be valuable
• C3: NEVER overfit conclusions to limited data — acknowledge sample limitations

Related

• SKILL-016: aget-analyze-data specification
• ONTOLOGY_analyst.yaml: Dataset, Analysis, Insight concepts
• CAP-ANL-001: Data Analysis capability