Reducing Overfitting — Applying L2 Regularization First
A healthcare startup is building an AI system to predict patient readmission risk. The team collects structured data from electronic health records (EHR) including age, diagnosis codes, lab results, and previous admissions. During initial training, the model achieves 95% accuracy on the validation set but only 60% accuracy on a holdout test set from a different hospital. The data scientist suspects overfitting. Which action should the team take first to improve generalization?
Quick Answer
The answer is to apply L2 regularization first, as this directly penalizes large weights by adding their squared magnitude to the loss function, forcing the model to learn simpler patterns that generalize beyond the training hospital’s data. This technique reduces variance—the core symptom when validation accuracy is high but test accuracy drops sharply—by discouraging the model from fitting noise in the electronic health records. On the CompTIA AI+ AI0-001 exam, this scenario tests your ability to distinguish overfitting from underfitting and to prioritize regularization over collecting more data or reducing model complexity, which are secondary steps. A common trap is choosing dropout or early stopping, but L2 regularization is the first-line defense because it smoothly shrinks coefficients without discarding features. Memory tip: think “L2 = Large weight Limiter”—it adds a penalty that keeps the model’s decision boundaries smooth and less wiggly, just like a ridge flattens a mountain range.
⚠ Common exam trap
CompTIA often tests the misconception that overfitting is always solved by more data, but the trap here is that collecting more data from the same source does not fix distribution shift—regularization directly penalizes model complexity to improve generalization to unseen distributions.
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
✓
Apply L2 regularization to the model
The model's high accuracy on the validation set but poor accuracy on a holdout test set from a different hospital indicates overfitting to the training data's specific patterns, which do not generalize to new data. L2 regularization (ridge regression) adds a penalty proportional to the square of the weights, discouraging the model from fitting noise and encouraging simpler, more generalizable decision boundaries. This directly addresses overfitting by reducing variance without requiring more data or reducing model capacity too drastically.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Apply L2 regularization to the model
Why this is correct
Regularization penalizes large coefficients, reducing overfitting and improving generalization to new data.
- ✗
Switch to a linear regression model
Why it's wrong here
Simplifying the model may help, but regularization is a more nuanced approach and typically the first step.
- ✗
Increase the model complexity by adding more layers
Why it's wrong here
Increasing complexity typically worsens overfitting, especially when the model already overfits.
- ✗
Collect more data from the same hospital
Why it's wrong here
More data from the same source may not address the distribution shift; it could reinforce existing biases.
About these practice questions
One of 754 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 →
Same concept, more angles
1 more way this is tested on AI0-001
These questions test the same concept from different angles. Work through them to make sure you can recognise it however the exam phrases it.
Variation 1. A data scientist notices the model overfits. Which change to the exhibit's configuration would most likely reduce overfitting?
hard- A.Remove dropout layers
- B.Increase learning rate to 0.01
- ✓ C.Add L2 regularization to dense layers
- D.Increase units in the first dense layer to 512
Why C: Adding L2 regularization to dense layers penalizes large weights by adding a squared magnitude term to the loss function, which forces the model to learn simpler patterns and reduces overfitting. This directly addresses the core issue of the model memorizing noise in the training data.
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.