A correlation matrix is a table of pairwise correlation values. For missing-data analysis, it compares binary indicators showing whether each column value is missing. section 2.5.2
Use the matrix to investigate whether missing values follow a pattern:
- Create missingness indicators with pandas, then calculate correlations between their columns. section 2.5.2
- Interpret high positive values as evidence that missing values tend to co-occur. section 2.5.2
- Inspect the result visually with a Seaborn heatmap, while checking the generated code and outputs. section 2.5.2
Missingness patterns
A heatmap can reveal that missing values in two product columns frequently coincide, suggesting a shared data-collection or recording pattern worth investigating. section 2.5.2
