# Data Profiler

> Profile a dataset by computing statistics, data types, distributions, and quality metrics. Use when asked to profile, describe, or summarize a dataset.

- Skill: `gu-dsan6725/data-profiler` (Agent Skill)
- Install (CLI): `npx skillmds@latest add gu-dsan6725/data-profiler`
- Raw SKILL.md: https://api.skillmd.com/api/skills/gu-dsan6725/data-profiler/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: gu-dsan6725 (https://skillmd.com/u/gu-dsan6725)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/gu-dsan6725/data-profiler

---


When profiling a dataset, generate a comprehensive data profile:

1. **Basic Info**: Number of rows, columns, memory usage
2. **Column Types**: Data type for each column, count of numeric vs categorical
3. **Summary Statistics**: Mean, median, std, min, max, quartiles for numeric columns
4. **Missing Values**: Count and percentage of nulls per column
5. **Unique Values**: Count of unique values per column
6. **Distribution Shape**: Skewness and kurtosis for numeric columns
7. **Outlier Detection**: Count of outliers per column using the IQR method
8. **Correlation**: Top 5 most correlated feature pairs

Save the profile as a markdown report to `output/data_profile.md`.

Use polars for data manipulation. Follow the project's coding standards.