Data Analysis1 / 5
A data analyst receives a large dataset with 200 features (columns) for building a classification model. Many features may be redundant or irrelevant. Which technique should the analyst use to reduce the number of features while retaining the most important information?
CorrectIncorrect
Alex
With 200 features, the analyst faces the "curse of dimensionality"—too many variables cause overfitting, longer training, and poor generalization. Dimensionality reduction addresses this via feature selection (removing irrelevant features using correlation or importance scores) or feature extraction like PCA, which creates new components capturing maximum variance in fewer dimensions (CompTIA Data+ Obj 3.2; NIST SP 800-188). Why not the others? Imputation fills missing values—it doesn't reduce feature count. Deduplication removes duplicate rows, not columns. Encryption secures data but doesn't change dimensionality. Exam tip: "too many features" = dimensionality reduction; "missing values" = imputation; "duplicate records" = deduplication.
Sourcescikit-learn.org
Follow-up answers are available in the app. Create a free account — no credit card required.
Question 1 of 5
Create a free account