If a natural order exists among categories, which encoding approach should be used?

Get ready for the GARP Risk and AI Exam with flashcards and multiple choice questions. Each question comes with hints and explanations. Prepare for success!

Multiple Choice

If a natural order exists among categories, which encoding approach should be used?

When categories have a natural ranking, you want to preserve that order in the model input. Ordinal encoding maps each category to a numeric value that reflects its position in the order (for example, low = 1, medium = 2, high = 3). This allows the model to learn that the difference between levels follows a sense of progression, which can improve predictive performance when the order matters. One-hot encoding would strip away any sense of ranking by treating each category as independent, and isn’t suitable when order carries meaning. Imputation is about filling in missing values, not encoding categories, and data scaling is for numeric features, not categorical ones. So, ordinal encoding is the appropriate approach for ordinal categories.

When categories have a natural ranking, you want to preserve that order in the model input. Ordinal encoding maps each category to a numeric value that reflects its position in the order (for example, low = 1, medium = 2, high = 3). This allows the model to learn that the difference between levels follows a sense of progression, which can improve predictive performance when the order matters. One-hot encoding would strip away any sense of ranking by treating each category as independent, and isn’t suitable when order carries meaning. Imputation is about filling in missing values, not encoding categories, and data scaling is for numeric features, not categorical ones. So, ordinal encoding is the appropriate approach for ordinal categories.

More practice questions