DY0-001: CompTIA DataAIDemo

Question

A data scientist is developing a model with a very large number of independent features. Only a small number of features are meaningful, but the data scientist does not have the domain expertise to choose the right subset. Which of the following algorithmic methods should the data scientist use to best decrease the number of features while preserving interpretability?

A Linear regression
B Logistic regression
C LASSO regression
D Ridge regression

Explanation

LASSO regression uses an L1 penalty that produces a sparse coefficient vector, setting many feature coefficients exactly to zero. This performs embedded feature selection while retaining an interpretable linear model based on the remaining nonzero coefficients.

Learn more

scikit-learn: Linear Models — Lasso

DY0-001: CompTIA DataAIDemo

AstroTutor

Notes

Readiness

QuestionQ126

Community Discussion