You have a tabular dataset with 150 correlated features that have different numeric ranges and you need to reduce dimensionality with minimal impact on model performance before training a regression model. Which preprocessing approach meets this goal?

The correct approach involves scaling the features before applying PCA. PCA is sensitive to the scale of the features because it aims to maximize variance. Features with larger numeric ranges can disproportionately influence the principal components. Min-Max scaling (or standardization) ensures all features contribute equally to the variance calculation. SageMaker Data Wrangler provides a convenient interface for these preprocessing steps. Applying PCA directly to raw data (first incorrect option) would lead to components dominated by features with larger scales. Removing highly correlated features (second incorrect option) before PCA is unnecessary as PCA inherently handles correlation by creating orthogonal components. Removing least correlated features (third incorrect option) is generally not an effective dimensionality reduction strategy and doesn't address the scaling issue for PCA.