Which preprocessing technique will result in the GREATEST increase of the model’s inference accuracy?
Normalize the problematic features.
Bootstrap the problematic features.
Remove the problematic features.
Extrapolate synthetic features.
Explanations:
Normalizing the features ensures they are on the same scale, which helps the model process the data more effectively and prevents the dominance of high-value features.
Bootstrapping is used for resampling the dataset, typically to increase training data, but it won’t address the issue of feature scaling or improve the model’s ability to handle large feature values.
Removing the features would eliminate potentially valuable information that could improve model performance, especially if the features are independently informative as the business expert mentioned.
Extrapolating synthetic features could introduce noise or irrelevant information, potentially making the model more complex and leading to worse performance, rather than improving accuracy.