missingness-profile

name	missingness-profile
description	Report null percentages and recommended imputations for water quality data. Use when profiling missing data, deciding imputation strategy, or assessing data completeness.
user-invocable	true
argument-hint	Path to data file or DataFrame variable name

Missingness Profile

Profile null/missing values and recommend imputation strategies.

When to Use

• Before model training or KPI computation
• When missing data percentage is unknown
• When deciding between imputation strategies

Procedure

1. Compute per-column null counts and percentages
2. Classify missingness severity:
   • Low (< 5%): safe to impute with median/mode
   • Medium (5–20%): impute with care; flag in output
   • High (> 20%): warn; consider dropping column or using forward-fill for time series
3. Recommend imputation method per column based on dtype and missingness pattern
4. Check for systematic missingness patterns (e.g., always null for a specific site_id)

Output

Column	Null %	Severity	Recommended Action
...	...	...	...

Plus a summary of any systematic patterns found.