PCA for Dimensionality Reduction

Medium

machine-learning Google Palantir Bloomberg

After running PCA on a 200-feature dataset, the first 15 principal components explain 95% of the total variance. What is the practical implication of retaining only these 15 components?

Multiple Choice

The remaining 185 components are noise and should always be discarded You reduce the feature space from 200 to 15 while retaining most information, but the components lose direct interpretability The model accuracy will always improve because fewer features prevent overfitting You must keep all 200 components to avoid losing any predictive signal

Correct!

Incorrect — try again next time!

Upgrade to Pro to unlock solutions and explanations

PCA for Dimensionality Reduction

Multiple Choice

Related problems

Deep-dive guides