mediumds-missing-data

How do you handle missing data and when is imputation risky?

Answer

First understand why data is missing (MCAR/MAR/MNAR). Options: - Drop rows/columns (if small impact) - Impute (mean/median, KNN, model-based) - Add missingness indicators Imputation is risky when missingness is informative (MNAR) or when it creates false confidence. Always validate downstream impact.

Related Topics

Data CleaningStatisticsData Science

Related Questions

What is the bias–variance tradeoff and how does it affect model performance?

How do you split data into train/validation/test sets and why does it matter?

What is cross-validation and when should you use it?

Back to Data Scientist All Professions