midProbability
What is covariance and how is it different from correlation?
Updated May 17, 2026
Short answer
Covariance measures directional relationship; correlation measures standardized strength of relationship.
Deep explanation
Covariance depends on scale and units, making interpretation difficult across datasets. Correlation normalizes covariance by dividing by standard deviations, producing a bounded value between -1 and 1, allowing comparison across variables.
Real-world example
Covariance between rainfall and crop yield; correlation helps compare across regions.
Common mistakes
- Assuming covariance magnitude is comparable across datasets.
Follow-up questions
- Can covariance be zero but correlation non-zero?
- Why is correlation bounded?