CPMAI_V7 practice question 28 of 35

You are working with a dataset that has a high number of dimensions. You're running into issues because some dimensions don't have enough real examples to properly train the systems for predictable results. What's your best course of action?

Full set: 100 questions

Answer options

A: Keep going as planned and the problem will eventually correct itself

B: Try to get additional data - at least 5 training examples for each dimension in the representation Correct answer

C: Try to get additional information from project lead to see how many examples per dimension are needed

D: Try to improve the quality of your data through more preparation

Correct Answer:

B. Try to get additional data - at least 5 training examples for each dimension in the representation

Explanation

With many dimensions, sample sparsity impairs learning (curse of dimensionality). Standard heuristics recommend securing at least 5–10 training instances per feature so parameter estimates become statistically reliable. Procuring additional data therefore directly addresses the dimensional sparsity problem.

Why Incorrect

A. Continuing unchanged leaves each feature under-represented and models unstable.
C. Managerial input can’t override mathematical sample-size needs.
D. Cleaning data cannot invent missing examples; quantity, not just quality, is lacking.

References

1. Domingos, P. “A Few Useful Things to Know About Machine Learning,” Commun. ACM 55(10), 2012, p. 83.

2. Stanford CS229 Lecture Notes on Sample Complexity, 2021, §2, p. 5 (rule-of-thumb ≥5×features).

Keep practicing

Take the free CPMAI_V7 practice test

Ten exam-style questions with answers and explanations, plus the exam facts and study guides.

Start the free practice test