AI0-001 AI Models and Data Engineering Practice Question
During feature engineering, a data scientist creates a new feature that is a linear combination of two existing features. What risk does this pose to the model?
⚠ Common exam trap
CompTIA often tests the distinction between multicollinearity and overfitting, trapping candidates who confuse feature redundancy with model complexity.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Multicollinearity
Creating a new feature as a linear combination of two existing features introduces perfect multicollinearity, where the new feature is an exact linear function of the original ones. This violates the assumption of no perfect multicollinearity in linear models, causing the design matrix to become singular and making coefficient estimates unstable or impossible to compute. Even in non-linear models, high multicollinearity can inflate variance and reduce interpretability.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Multicollinearity
Why this is correct
A feature built as a linear combination of two existing features is perfectly correlated with them, producing multicollinearity. This inflates the variance of coefficient estimates, making them unstable and hard to interpret, which is the specific risk the engineered linear combination introduces.
- ✗
Data leakage
Why it's wrong here
Data leakage means training on information unavailable at prediction time, such as target-derived or future data; combining two existing input features introduces no such leakage. It is tempting because leakage is a common feature-engineering pitfall, and would be correct if the new feature were computed using the label or post-outcome data.
- ✗
Overfitting
Why it's wrong here
The derived feature is perfectly collinear with its two parents, adding no new information; the real risk is multicollinearity destabilising coefficient estimates, not overfitting, which requires excess model capacity relative to data. It is tempting because redundant features are often loosely blamed for overfitting, but the mechanism here is linear dependence.
- ✗
Underfitting
Why it's wrong here
A linear combination of existing features adds no new information, so it cannot cause underfitting — underfitting arises when the model is too simple to capture the underlying relationship in the data. It is tempting because redundant features are sometimes associated with poor model fit, but the actual risk here is multicollinearity.
About these practice questions
One of 962 original AI0-001 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.