Training a model step by step quiz

Question 1

What is the very first step before you can train a machine learning model?

Accepted Answer

Collect and prepare data

Answer

Buy a faster computer

Answer

Pick a colorful chart

Answer

Write the final report

Question 2

In simple terms, what does it mean to 'train' a model?

Accepted Answer

Adjusting its parameters using examples to improve predictions

Answer

Deleting all the data

Answer

Drawing the model by hand

Answer

Turning the computer off and on

Question 3

What is a 'feature' in a dataset?

Accepted Answer

An input attribute describing each example

Answer

A bug in the code

Answer

The model's final score

Answer

The name of the computer

Question 4

What is a 'label' in supervised learning?

Accepted Answer

The correct answer attached to an example

Answer

A sticker on the laptop

Answer

A type of graph

Answer

The speed of training

Question 5

Why do we split data into training and testing sets?

Accepted Answer

To test the model on data it has never seen

Answer

To make training slower

Answer

To delete half the data

Answer

To save battery power

Question 6

What is a 'loss function' used for during training?

Accepted Answer

It measures how wrong the predictions are

Answer

It counts how many files were lost

Answer

It picks the screen color

Answer

It deletes wrong data

Question 7

What does the model try to do with the loss value while training?

Accepted Answer

Minimize it

Answer

Maximize it

Answer

Ignore it

Answer

Keep it constant

Question 8

What is an 'epoch' in model training?

Accepted Answer

One full pass through all the training data

Answer

One single example

Answer

One mistake by the model

Answer

One hour of training

Question 9

What is the 'learning rate' in training?

Accepted Answer

How big a step the model takes when updating

Answer

How fast the student learns math

Answer

The number of examples

Answer

The size of the screen

Question 10

What is 'overfitting'?

Accepted Answer

Memorizing training data and failing on new data

Answer

Training too few examples

Answer

Using a small screen

Answer

Forgetting all the data

Question 11

What is 'underfitting'?

Accepted Answer

The model is too simple and learns too little

Answer

The model memorizes everything

Answer

The data is too clean

Answer

The computer is too fast

Question 12

Why is it important to clean your data before training?

Accepted Answer

Messy data can lead to wrong predictions

Answer

Cleaning makes the file larger

Answer

It changes the screen color

Answer

It is required by law

Question 13

What is a 'validation set' used for?

Accepted Answer

To check progress and tune settings during training

Answer

To store backup photos

Answer

To speed up the computer

Answer

To label the data automatically

Question 14

What does 'accuracy' tell us about a classification model?

Accepted Answer

The fraction of predictions that are correct

Answer

The speed of the model

Answer

The number of features

Answer

The size of the dataset

Question 15

In the training loop, what happens after the model makes a prediction?

Accepted Answer

It compares the guess to the answer and updates parameters

Answer

It deletes the prediction

Answer

It restarts the computer

Answer

It saves the photo

Question 16

What is 'gradient descent' in simple words?

Accepted Answer

A method that adjusts parameters to lower the loss

Answer

A way to draw graphs

Answer

A type of dataset

Answer

A name for the screen

Question 17

Why should the test data be kept separate and not used during training?

Accepted Answer

To get an honest measure on unseen data

Answer

To make the file smaller

Answer

To save electricity

Answer

To make training faster

Question 18

What is a simple way to use scikit-learn to split data in Python?

Accepted Answer

train_test_split

Answer

delete_data

Answer

color_picker

Answer

make_loud

Question 19

What is a 'parameter' that a model learns during training?

Accepted Answer

An internal number the model adjusts to fit data

Answer

The name of the file

Answer

The screen brightness

Answer

The number of users

Question 20

What is the difference between a parameter and a hyperparameter?

Accepted Answer

Parameters are learned; hyperparameters are set before training

Answer

They are exactly the same thing

Answer

Both are pictures

Answer

Both are set after training

Question 21

Why might you train a model for more than one epoch?

Accepted Answer

The model keeps improving with more passes

Answer

More epochs delete data

Answer

One epoch erases the model

Answer

Epochs change the screen color

Question 22

What does 'normalizing' or 'scaling' features mean?

Accepted Answer

Putting features on a similar range

Answer

Deleting half the features

Answer

Making the data colorful

Answer

Counting the rows

Question 23

What is 'batch size' in training?

Accepted Answer

Number of examples seen before each update

Answer

The size of the hard drive

Answer

The number of epochs

Answer

The screen resolution

Question 24

How can you tell if your model is overfitting during training?

Accepted Answer

Training score rises while validation score drops

Answer

Both scores reach exactly zero

Answer

The computer overheats

Answer

The dataset disappears

Question 25

What is the goal of 'generalization' in machine learning?

Accepted Answer

Performing well on new, unseen data

Answer

Memorizing the training data only

Answer

Running on a big screen

Answer

Using less memory

Question 26

Why is having a large and varied dataset usually helpful for training?

Accepted Answer

It helps the model learn patterns that generalize

Answer

It always makes training instant

Answer

It removes the need for testing

Answer

It changes the colors

Question 27

What is an ethical problem if your training data is biased?

Accepted Answer

The model can make unfair decisions

Answer

The model trains faster

Answer

The screen turns red

Answer

The data becomes free

Question 28

After training, what is the next step to check if the model is good?

Accepted Answer

Evaluate it on the test set

Answer

Delete the model

Answer

Train it on the same data forever

Answer

Turn off the computer

Question 29

In scikit-learn, what method is commonly used to train a model on data?

Accepted Answer

fit

Answer

color

Answer

close

Answer

delete

Question 30

What does it mean to 'tune' a model's hyperparameters?

Accepted Answer

Trying different settings to get the best results

Answer

Painting the model

Answer

Deleting the dataset

Answer

Restarting training from zero each time

Test yourself