Skip to content

FRM Part I · FRM Exam Part I · Machine-Learning Methods

A modeler has a strongly right-skewed feature, transaction size, with values ranging from 100 to 10,000,000 dollars. Extreme values are valid, not errors. To reduce the influence of the skew in a linear model, which treatment is MOST appropriate?

Apply a logarithmic transformation. It compresses very large values and reduces right skew while keeping valid extreme observations. Deleting them discards real information, squaring would worsen the skew, and replacing values with the mean would eliminate the feature's predictive content.

  1. AApply a logarithmic transformation to the featureCorrect
  2. BDelete all observations above the 99th percentile as errors
  3. CSquare the feature to emphasize large values
  4. DReplace every value with the sample mean

Explanation

A log transformation compresses large values and reduces right skew while retaining all valid observations. Deleting extremes discards valid information, squaring increases skew, and replacing values with the mean removes all information in the feature.

Did you get it right without looking?

One question tells you little. A timed set on Machine-Learning Methods shows your real accuracy, how long you take and where you lose marks.

Practise Machine-Learning MethodsFree diagnostic mock

More Machine-Learning Methods questions