| name | csv-eda-basica |
| description | Use when starting exploratory data analysis on a CSV file. Detects columns, types, missing values, and distributions. Proposes initial questions and produces a structured EDA notebook. Invoke with: /csv-eda-basica, eda, explore dataset, analyze csv, check data quality. |
| argument-hint | Path to the CSV file to analyse (e.g. data/sample.csv) |
| user-invocable | true |
Skill: csv-eda-basica
Perform a structured Exploratory Data Analysis (EDA) on a CSV file.
Steps
1. Load and Inspect
- Read the CSV at the path provided in the argument.
- Report: shape, column names, data types, first 5 rows.
2. Data Quality Check
- Count missing values per column (absolute and percentage).
- Identify columns with >20% missing values and flag them as high-risk.
- Check for duplicated rows.
- Refer to the checklist in
references/eda-checklist.md.
3. Distributions
- For numeric columns: mean, std, min, max, quartiles.
- For categorical columns: value counts and top-5 categories.
- Note any columns with suspiciously low variance (potential constants).
4. Propose Initial Questions
- Based on the structure and distributions found, list 3–5 analytical questions worth investigating.
- Example: "Is there a correlation between feature X and the target variable?"
5. Generate Notebook
- Use the notebook template at
assets/notebook-template.md as the structure.
- Create a new
.ipynb file at notebooks/eda_<dataset_name>.ipynb with cells matching each step above.
- Follow conventions in
.github/instructions/notebooks.instructions.md.
Output
A populated EDA notebook ready to run, plus a brief summary in the chat of key findings.