analyzing-data

# Data Analysis

Answer business questions by querying the data warehouse. The kernel starts automatically on first use.

## Prerequisites

**uv must be installed:**
```bash
curl -LsSf https://astral.sh/uv/install.sh | sh
```

Scripts are located relative to this skill file.

## MANDATORY FIRST STEP

**Before any other action, check for cached patterns:**

```bash
uv run scripts/cli.py pattern lookup "<user's question>"
```

This is NON-NEGOTIABLE. Patterns contain proven strategies that save time and avoid failed queries.

---

## Workflow

```
Analysis Progress:
- [ ] Step 1: pattern lookup (check for cached strategy)
- [ ] Step 2: concept lookup (check for known tables)
- [ ] Step 3: Search codebase for table definitions (Grep)
- [ ] Step 4: Read SQL file to get table/column names
- [ ] Step 5: Execute query via kernel (run_sql)
- [ ] Step 6: learn_concept (ALWAYS before presenting results)
- [ ] Step 7: learn_pattern (ALWAYS if discovery required)
- [ ] Step 8: record_pattern_outcome (if you used a pattern in Step 1)
- [ ] Step 9: Present findings to user
```

---

## CLI Commands

### Kernel Management

```bash
uv run scripts/cli.py start           # Start kernel with Snowflake
uv run scripts/cli.py exec "..."      # Execute Python code
uv run scripts/cli.py status          # Check kernel status
uv run scripts/cli.py restart         # Restart kernel
uv run scripts/cli.py stop            # Stop kernel
uv run scripts/cli.py install plotly  # Install additional packages
```

### Concept Cache (concept -> table mappings)

```bash
# Look up a concept
uv run scripts/cli.py concept lookup customers

# Learn a new concept
uv run scripts/cli.py concept learn customers HQ.MART_CUST.CURRENT_ASTRO_CUSTS -k ACCT_ID

# List all concepts
uv run scripts/cli.py concept list

# Import concepts from warehouse.md
uv run scripts/cli.py concept import -p /path/to/warehouse.md
```

### Pattern Cache (query strategies)

```bash
# Look up patterns for a question
uv run scripts/cli.py p