Machine Learning Fundamentals quiz

Question 1

What is the key difference between supervised and unsupervised learning?

Accepted Answer

Supervised uses labeled data to predict outputs; unsupervised finds patterns in unlabeled data

Answer

Supervised uses no data; unsupervised uses all data

Answer

Supervised is only for images; unsupervised is only for text

Answer

Supervised needs no training; unsupervised needs training

Question 2

Why do we split a dataset into training and test sets?

Accepted Answer

To estimate how well the model generalizes to unseen data

Answer

To make training faster by using less data

Answer

Because algorithms require exactly two files

Answer

To increase the number of features available

Question 3

What is overfitting in machine learning?

Accepted Answer

Model fits training data including noise and fails on new data

Answer

Model is too simple to learn anything

Answer

Model trained without any data

Answer

Model that always predicts the average value

Question 4

What is underfitting and what typically causes it?

Accepted Answer

A too-simple model that performs poorly on both training and test data

Answer

A model that perfectly fits the test set

Answer

A model that overfits the noise

Answer

A model with too many parameters

Question 5

Is classification a supervised or unsupervised task, and what does it predict?

Accepted Answer

Supervised; it predicts a discrete category label

Answer

Unsupervised; it predicts a continuous value

Answer

Supervised; it predicts a continuous value

Answer

Unsupervised; it groups data without labels

Question 6

How does regression differ from classification?

Accepted Answer

Regression predicts continuous values; classification predicts categories

Answer

Regression predicts categories; classification predicts numbers

Answer

Both predict only categories

Answer

Both are unsupervised methods

Question 7

What is a 'feature' in a machine learning dataset?

Accepted Answer

An input variable describing each example

Answer

The final prediction of the model

Answer

The error of the model

Answer

The learning rate setting

Question 8

What does the 'label' or 'target' mean in supervised learning?

Accepted Answer

The correct output the model learns to predict

Answer

A random input feature

Answer

The number of training examples

Answer

The model's internal weight

Question 9

What is the goal of clustering algorithms like k-means?

Accepted Answer

Group similar data points together without using labels

Answer

Predict a numeric label for each point

Answer

Reduce the number of training epochs

Answer

Label data using a teacher signal

Question 10

In k-means clustering, what does the parameter k represent?

Accepted Answer

The number of clusters to form

Answer

The number of features in the data

Answer

The number of training iterations only

Answer

The learning rate of the model

Question 11

What is the purpose of a loss (or cost) function?

Accepted Answer

It measures prediction error so training can minimize it

Answer

It counts the number of features

Answer

It stores the dataset in memory

Answer

It selects which algorithm to use automatically

Question 12

What is gradient descent used for in training models?

Accepted Answer

To minimize the loss by iteratively updating parameters

Answer

To split data into train and test sets

Answer

To label unlabeled data

Answer

To increase the model's loss on purpose

Question 13

What does the learning rate control in gradient descent?

Accepted Answer

The size of each parameter update step

Answer

The number of features used

Answer

The number of classes to predict

Answer

The amount of training data used

Question 14

What is the difference between a model parameter and a hyperparameter?

Accepted Answer

Parameters are learned from data; hyperparameters are set before training

Answer

Both are learned automatically from data

Answer

Both are set manually before training

Answer

Parameters are set by the user; hyperparameters are learned

Question 15

What is the bias-variance tradeoff?

Accepted Answer

The balance between underfitting (high bias) and overfitting (high variance)

Answer

The choice between supervised and unsupervised learning

Answer

The ratio of training to test data

Answer

The tradeoff between speed and memory

Question 16

What is the role of a validation set, separate from the test set?

Accepted Answer

To tune hyperparameters without touching the final test set

Answer

To replace the training data entirely

Answer

To store labels for unsupervised learning

Answer

To increase the number of features

Question 17

What is k-fold cross-validation and why is it useful?

Accepted Answer

Splitting data into k parts to average performance over multiple train/validation splits

Answer

Training the model k times on the same fixed split

Answer

Using k features instead of all features

Answer

Choosing k clusters in unsupervised learning

Question 18

In a confusion matrix, what is a 'false positive'?

Accepted Answer

Predicting positive when the true label is negative

Answer

Predicting negative when the true label is negative

Answer

Predicting positive when the true label is positive

Answer

Predicting negative when the true label is positive

Question 19

How is accuracy defined for a classification model?

Accepted Answer

Correct predictions divided by total predictions

Answer

Total predictions divided by correct predictions

Answer

The number of features used

Answer

The model's training time

Question 20

Why can accuracy be a poor metric for imbalanced datasets?

Accepted Answer

A model can score high by always predicting the majority class

Answer

Accuracy cannot be computed for imbalanced data

Answer

Accuracy always equals precision

Answer

Imbalanced data makes accuracy negative

Question 21

What do precision and recall measure in classification?

Accepted Answer

Precision = correct positives among predicted positives; recall = correct positives among actual positives

Answer

Precision and recall both measure total accuracy

Answer

Precision measures speed; recall measures memory

Answer

They both count the number of features

Question 22

What is the F1 score and when is it preferred?

Accepted Answer

The harmonic mean of precision and recall

Answer

The average of accuracy and precision

Answer

The total count of correct predictions

Answer

The difference between precision and recall

Question 23

What problem does feature scaling (normalization/standardization) solve?

Accepted Answer

It prevents large-valued features from dominating and helps convergence

Answer

It adds new labels to the dataset

Answer

It removes the need for a test set

Answer

It converts regression into classification

Question 24

How does the k-nearest neighbors (KNN) algorithm classify a new point?

Accepted Answer

By taking a majority vote of its k closest training points

Answer

By averaging all training labels

Answer

By fitting a straight line through the data

Answer

By randomly assigning a class

Question 25

What is the basic idea behind a decision tree?

Accepted Answer

It splits data with a series of feature-based questions leading to predictions

Answer

It clusters data into k groups

Answer

It always fits a straight line

Answer

It memorizes the test set

Question 26

What is a random forest and why does it often outperform a single decision tree?

Accepted Answer

It combines many decision trees to reduce overfitting and variance

Answer

It is a single very deep decision tree

Answer

It is an unsupervised clustering method

Answer

It only works on image data

Question 27

Despite its name, what type of task is logistic regression used for?

Accepted Answer

Classification, by outputting class probabilities

Answer

Predicting continuous numeric values

Answer

Clustering unlabeled data

Answer

Reducing dimensionality

Question 28

What does the sigmoid function output and why is it useful in classification?

Accepted Answer

A value between 0 and 1, interpretable as a probability

Answer

A value between -1 and 1 only

Answer

Any integer value

Answer

Always exactly 0 or 1

Question 29

What is the equation form of simple linear regression?

Accepted Answer

y = mx + b, a straight line with slope and intercept

Answer

y = x squared

Answer

y = log(x)

Answer

y is unrelated to x

Question 30

What does mean squared error (MSE) measure in regression?

Accepted Answer

The average of squared differences between predictions and actual values

Answer

The number of correct classifications

Answer

The average of the raw prediction values

Answer

The maximum error in the dataset

Test yourself