What is the purpose of normalization in machine learning data preprocessing?

Correct answer: Scale features to a similar range to improve model convergence and performance

Option A is correct because normalization (and the related technique of standardization) rescales numerical features to a common range or distribution, which prevents features with large magnitudes from dominating gradient updates, speeds up convergence of optimization algorithms such as gradient descent, and improves the stability and accuracy of many models. Option B is wrong because encoding categorical variables (such as one-hot encoding or label encoding) is a distinct preprocessing step that converts non-numeric categories into numeric representations, not a normalization operation. Option C is incorrect because handling missing values involves imputation strategies or removal of null entries, which is a separate data-cleaning step performed independently of scaling. Option D is wrong because removing outliers is an optional data-cleaning technique that addresses extreme values, distinct from normalization which scales the range of all values.