wayground logo

Free Printable Worksheets

NEW

Font size

S
M
L
XL
Worksheets

ML_quiz_redi

Total questions: 23

Worksheet time: 12mins

Name
Class
Date
1.
Which phrase best defines ML?
a)
Computers learn from data
b)
Manually coded rules
c)
Designing hardware
d)
Creating web apps
2.
What is needed for ML to work well?
a)
dataset representative of the target use case
b)
No training needed
c)
Random guesses
d)
Manual code rules
3.
Which is NOT part of an ML workflow?
a)
Feature engineering
b)
Hyperparam tuning
c)
Manual coding all rules
d)
Model validation
4.

Why did machine learning only become practical in the last few decades, even though the core ideas existed earlier?

a)

Lack of data and computational resources

b)

Lack of programming languages

c)

ML wasn’t mathematically possible until the 2000s

d)

Humans weren’t interested in automation

5.
What is hyperparameter tuning?
a)
Changing classification to regression
b)
Adjusting model weights after training
c)
Randomly removing large parts of your dataset
d)
Optimizing parameters that aren’t learned directly (e.g., learning rate)
6.
How can you ensure that your model isn't overfitting?
a)
Use a separate validation/test set or perform cross-validation
b)
Remove data from the training set randomly
c)
Keep increasing your model’s complexity
d)
Ignore performance metrics on unseen data
7.
What is NOT a principle of responsible AI?
a)
Inclusiveness
b)
Transparency
c)
Accountability
d)
Operations
8.
Why do we do Exploratory Data Analysis?
a)
To choose final model
b)
To test production code
c)
To understand data better
d)
To improve GPU usage
9.
Which library is used for data visualization?
a)
Matplotlib
b)
VS Code
c)
grep terminal command
d)
MS Paint
10.
Why use a box plot in EDA?
a)
Shows correlation
b)
Shows freq distribution
c)
Shows median & outliers
d)
Shows time-series trend
11.

Why is it important to visualize data before fitting a regression model?

a)

To check the beauty of the dataset

b)

To detect relationships, outliers, and non-linear trends

c)

To reduce training time

d)

To automatically select features

12.
What does regression predict?
a)
Yes/No labels
b)
Continuous values
c)
Document clusters
d)
Chess moves
13.
Which is a typical regression use?
a)
Forecasting house prices
b)
Email spam detection
c)
Clustering customers
d)
Text translation
14.
Which metric is NOT for regression?
a)
MSE
b)
R-squared
c)
Accuracy
d)
MAE
15.

Which of the following is a risk when increasing the polynomial degree in regression?

a)

Underfitting

b)

Perfect generalization

c)

Overfitting

d)

Improved interpretability

16.
Which features can be used without transformation for linear regression?
a)
numerical
b)
nominal (categorical)
c)
ordinal
d)
none of them
17.
Which method can be used to convert a nomial (categorical) feature into a numerical feature?
a)
normalization
b)
One hot encoding
c)
feature scaling
d)
power transform
18.
Key difference of polynomial regression to linear regression?
a)
Uses polynomial terms
b)
Simpler than linear
c)
Doesn't need data
d)
Only for images
19.
What is logistic regression used for?
a)
Regression
b)
Classification
c)
Data clustering
d)
Recommending products
20.
Which function maps input to (0;1)?
a)
Linear
b)
Sigmoid
c)
Exponential
d)
ReLU
21.
How interpret logistic reg coefficients?
a)
Linear prob changes
b)
Change in log-odds
c)
No direct interpretation
d)
They show predicted label
22.

In what scenario would you prefer a high recall model over a high precision one?

a)

Spam email filtering

b)

Medical diagnosis of a rare but serious disease

c)

Movie recommendation systems

d)

Predicting daily weather temperature

23.

What describe best gradient descent?

a)

The weights move randomly with high temperature and slowly stop moving as temperature decreases

b)

The slope starts vertically and then slowly goes down until a lowest loss is found

c)

Weights are moving in the direction of the partial derivative of the loss toward a local minimum

d)

This is an exact solver to find the global minimum of the loss