How does using a test dataset help prevent overfitting?

It automatically selects the best model architecture.

It provides a way to detect when the model performs well on training data but poorly on new data.

It reduces the complexity of the model.

It increases the amount of training data available.

A data analyst is building a model to predict customer satisfaction (binary). They have a dataset with 10,000 rows and 50 features. They split the data into training and test sets, train a decision tree, and achieve 85% accuracy on the test set. However, they suspect the model may be overfitting. Which approach is most appropriate to diagnose overfitting?

Increase the number of features.

Compare training accuracy to test accuracy.

Train the model on the test set.

Python InstitutePCAD - Certified Associate in Data Analytics with Python

Domain 4Objective 2

Statistical Methods and Machine Learning PCAD-31 Practice Questions (Page 1)

Part of the Data Analysis and Modeling domain, which accounts for 18.8% of the PCAD-31 exam. Python Institute does not publish an official question count, but from its 60-minute exam (~25–40 total, ~5–5 in this domain), expect 3–3 from this objective — we provide 30 practice questions to prepare you well beyond it. (estimate)

30questions here

6free pages

9concepts

18.8%of the exam

Questions 1–5

1application · medium
A model predicts whether a transaction is fraudulent. In the test set of 1,000 transactions, 20 are actually fraudulent. The model flags 30 transactions as fraudulent, of which 15 are truly fraudulent. What is the model's precision?
Select an answer first
2application · medium
A binary classifier is evaluated on a test set. The confusion matrix shows: True Positives = 40, False Positives = 10, True Negatives = 100, False Negatives = 20. What is the F1-score?
Select an answer first
3foundation · easy
Which supervised learning algorithm is most appropriate for predicting a continuous numeric value, such as house prices?
Select an answer first
4foundation · easy
How does using a test dataset help prevent overfitting?
Select an answer first
5expert · easy
A data analyst is building a model to predict customer satisfaction (binary). They have a dataset with 10,000 rows and 50 features. They split the data into training and test sets, train a decision tree, and achieve 85% accuracy on the test set. However, they suspect the model may be overfitting. Which approach is most appropriate to diagnose overfitting?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to PCAD-31

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Python Institute. “PCAD-31” is a trademark of its owner, used for identification only.

Statistical Methods and Machine Learning PCAD-31 Practice Questions (Page 1)

Questions 1–5

A model predicts whether a transaction is fraudulent. In the test set of 1,000 transactions, 20 are actually fraudulent. The model flags 30 transactions as fraudulent, of which 15 are truly fraudulent. What is the model's precision?

A binary classifier is evaluated on a test set. The confusion matrix shows: True Positives = 40, False Positives = 10, True Negatives = 100, False Negatives = 20. What is the F1-score?

Which supervised learning algorithm is most appropriate for predicting a continuous numeric value, such as house prices?

How does using a test dataset help prevent overfitting?

Keep studying — Data Analysis and Modeling