Replacing missing values in a dataset with estimated values based on other available data.
When a dataset has missing values, analysts must decide how to handle them. Imputation fills in the gaps with plausible estimates rather than leaving them null or dropping the records entirely. Common imputation methods include replacing missing values with the column mean or median, using the most frequent value, or applying a statistical model to predict the missing value from other columns.
Example: A survey dataset has 8% of income values missing. Rather than excluding those respondents from analysis, an analyst imputes the missing values using the median income for respondents with similar age, education, and region. The imputed values are flagged so users know they are estimates.
Imputation should always be documented. Users of the data need to know which values were imputed, what method was used, and how imputation might affect analysis results.