Free Practice Quiz Question List

8 Introduction to Machine Learning Online Quiz Questions

Use this free practice quiz with 20 questions to review 8 Introduction to Machine Learning, test your knowledge, and prepare for your next test or exam.

20 questions

Choose one

1 point

A model uses labeled examples to predict whether each new email is spam or not spam. Which type of prediction task is this?

A
Regression
B
Classification
C
Clustering
D
Unsupervised learning

True or false

1 point

True or false: If an unsupervised clustering algorithm groups customers with similar purchasing patterns, those groups are automatically meaningful customer categories.

A
True
B
False

Fill in the blank

1 point

A dataset is organized with examples in rows and in columns. What term completes the statement?

Choose one

1 point

A company wants to predict next month's demand using historical records. Which train-test split is most appropriate for estimating performance on future demand?

A
Randomly mix all dates so each set has a similar number of records.
B
Train on later dates and test on earlier dates.
C
Train on earlier dates and test on later dates.
D
Use the test set to select the model, then report its score as a final evaluation.

Choose all

1 point

Select all correct statements about the roles of training, validation, and test data.

A
Training data is used to fit the model.
B
A test set should be repeatedly used to guide model choices during development.
C
A validation set may be used to compare models or tune choices.
D
A test set is used to fit the model parameters.

Written response

1 point

Complete the statement: A model that performs well on new examples ___.

True or false

1 point

True or false: Repeatedly checking test-set results to guide model-development decisions can make the test set less independent as a final evaluation.

A
True
B
False

Choose one

1 point

A model is intended to predict whether a customer will renew a subscription next month. Which situation is an example of data leakage?

A
Using a validation set to compare candidate models.
B
Including in training a feature that reveals an outcome only recorded after the prediction must be made.
C
Testing on records from a later time period.
D
Using examples with known labels to train a supervised model.

Fill in the blank

1 point

In a screening task, missing a true case is especially costly. The team therefore pays particular attention to , which measures the fraction of actual positive cases detected.

Choose all

1 point

Select all correct statements about regression evaluation measures.

A
Mean squared error penalizes large errors more heavily because it averages squared errors.
B
Mean absolute error is the square root of mean squared error.
C
Root mean squared error is expressed in the target's units.
D
A lower value of any regression error measure always proves the model is suitable for every practical purpose.

Written response

1 point

A classifier has 36 true positives and 4 false positives. Using $\text{precision}=\frac{TP}{TP+FP}$ , what is its precision as a percentage? Enter the percentage value without the percent sign.

Choose one

1 point

A model fits its training examples very closely but performs poorly on a separate set of unseen examples. Which problem does this describe?

A
Underfitting
B
Generalization
C
Overfitting
D
Data leakage

Open ended

1 point

How can a clustering result be evaluated when there are no known labels, and how can the evaluation change if labels are available? Explain why a numerical score alone may not be enough.

True or false

1 point

True or false: A good test-set score guarantees that a model will perform well in every real-world setting.

A
True
B
False

Choose one

1 point

A company trains a model on emails labeled as spam or not spam and uses it to categorize new emails. Which kind of prediction task is this?

A
Regression, because the model predicts a value from input data
B
Classification, because the model predicts a category
C
Clustering, because the model predicts a known label
D
Unsupervised learning, because the model predicts a category

Written response

1 point

Which regression measure averages the absolute differences between predicted and actual values?

Choose one

1 point

A clustering algorithm groups customers with similar purchasing patterns, but no target labels were provided. What can the company reasonably conclude from the result?

A
The groups are known labels that the algorithm learned to predict
B
The groups must correspond to meaningful customer categories
C
The groups are patterns in the features that need interpretation
D
The groups prove that the customers have different causes for their purchases

Choose one

1 point

A screening system is used to flag people who may have a condition, and missing a true case is especially costly. Which classification measure should the team give particular attention?

A
Precision, because it measures how many predicted positives are truly positive
B
Recall, because it measures how many actual positive cases are detected
C
Accuracy, because it measures the fraction of all predictions that are correct
D
F1 score, because it measures only the cost of missed positive cases

Choose one

1 point

A retailer wants to predict demand for future weeks using historical records. Which split is most appropriate for estimating performance in the intended use?

A
Train on earlier dates and test on later dates to reflect the model’s intended use.
B
Randomly mix all dates so the training and test sets have similar time ranges.
C
Train on later dates and test on earlier dates.
D
Use only the earliest dates for both training and testing.

Written response

1 point

A model predicting whether a loan will default uses a training feature that is only recorded after the default occurs. What is this problem called?