Free Practice Quiz Question List

4 Machine Learning Basics Online Quiz Questions

Use this free practice quiz with 20 questions to review 4 Machine Learning Basics, test your knowledge, and prepare for your next test or exam.

20 questions

True or false

1 point

True or false: Repeatedly checking test-set performance while choosing a model can make the final test estimate less trustworthy.

A
True
B
False

True or false

1 point

True or false: A model that predicts whether a transaction is fraudulent is performing regression.

A
True
B
False

Written response

1 point

A supervised model predicts the numerical delivery time for each package. What type of learning task is this?

Fill in the blank

1 point

In supervised learning, examples pair input features with known that the model learns to predict.

Written response

1 point

What is the term for settings such as a learning rate that are chosen outside the model-fitting process?

Fill in the blank

1 point

When a model performs strongly on training examples but more weakly on unseen examples because it learned training-specific details or noise, it is experiencing .

Open ended

1 point

A delivery company predicts arrival times from historical shipments and expects seasonal patterns. Describe how it should split the data over time, use validation and test sets, select evaluation metrics, and respond if real-world data or outcomes change after deployment.

Choose one

1 point

A bank trains a model using past transactions labeled as fraudulent or legitimate, then uses it to assign one of those categories to new transactions. What type of task is this?

A
Regression
B
Classification
C
Clustering
D
Optimization

Choose one

1 point

A retailer has customer purchase data but no target labels. It wants to find groups of customers with similar purchasing patterns. Which approach best matches this goal?

A
Train a classifier using customer labels for preferred product categories
B
Predict each customer's spending amount from labeled purchase histories
C
Group customers by similarities in purchasing patterns without target labels
D
Assign each customer a known target label before fitting a model

True or false

1 point

For a time-dependent prediction task, evaluating on observations from later dates is generally more appropriate than randomly mixing past and future observations.

A
True
B
False

Written response

1 point

A model uses labeled examples to estimate a home's sale price, which is a numerical quantity. What type of supervised task is this?

Choose one

1 point

A model performs very well on its training examples but substantially worse on new examples. Which problem does this pattern most directly suggest?

A
The model has overfit details specific to its training examples
B
The model is underfitting because it performs well on training data
C
The model has demonstrated reliable generalization
D
The model is necessarily affected by distribution shift

Choose one

1 point

A team compares several learning-rate settings using validation data before choosing a model configuration. What role does the learning rate play in this process?

A
It is a model weight learned directly from each training example
B
It is the final test metric used to report model performance
C
It is a training setting chosen outside parameter fitting and typically tuned with validation data
D
It is a target label paired with an input feature

Written response

1 point

A team calculates imputation values using the full dataset before splitting it into training and test sets. What is the standard name for the resulting evaluation problem?

Choose one

1 point

A team needs to compare candidate models and then estimate how the selected model may perform on new examples. Which procedure best preserves a final independent assessment?

A
Use the test set repeatedly to choose model settings, then report the best result
B
Compare and tune candidate models on validation data, then use held-out test data for a final estimate
C
Use training data for the final estimate because it contains the most examples
D
Combine validation and test data before tuning so the comparison is more reliable

Choose one

1 point

A screening model must detect a rare but important condition. Why might overall accuracy alone be inadequate for evaluating it?

A
Accuracy always reveals whether every class is detected well
B
A rare class can be ignored if the overall accuracy is high
C
Use only the training loss because it directly measures prediction errors
D
Choose metrics that reflect the task and consequences of different errors

Choose one

1 point

A model's performance estimate was strong on a test set, but its predictions become less reliable months after deployment. Which concern most directly explains why ongoing monitoring matters?

A
Real-world inputs or outcomes may have changed, making earlier evaluation less representative
B
Monitoring guarantees that the model will not overfit its training data
C
A model's parameters automatically update whenever the input distribution changes
D
A change in real-world data proves that the original test set was incorrectly labeled

Choose one

1 point

A delivery company is considering a model to help dispatchers plan routes. Before gathering data, which step best defines the machine-learning task?

A
Start collecting every available field and decide what to predict after training several models.
B
Specify the prediction target, who will use the prediction, and how success will be measured.
C
Choose the model with the lowest training error before deciding how its predictions will be used.
D
Collect only examples that are easy to label, then define success as the number of examples collected.

Choose all

1 point

A clinic proposes using whether a patient returned for another appointment as a label for whether treatment was successful. Which checks would help determine whether this label is suitable? Select all that apply.

A
Check whether the label definition actually corresponds to the outcome the project wants to predict.
B
Assume that labels are accurate whenever they are present in a structured database.
C
Review examples for mistakes in how labels were assigned.
D
Consider whether the label is only an imperfect proxy for the outcome of interest.
E
Replace every label with the most common value so the model has a consistent target.

Choose all

1 point

A retailer is assembling customer data for a model to support service decisions. Which checks are appropriate when inspecting whether the examples are suitable? Select all that apply.

A
Check whether measurements were recorded correctly.
B
Look for duplicate examples that could distort the dataset.
C
Check whether the examples adequately represent the population where the model will be used.
D
Assume that a large dataset must be representative, regardless of how it was collected.
E
Remove every example from a small subgroup so the groups have equal counts.