Grouping data with clustering quiz

Question 1

What is clustering in machine learning?

Accepted Answer

Grouping similar data points together without pre-assigned labels

Answer

Teaching a computer to predict tomorrow's weather from past data

Answer

Sorting a list of numbers from smallest to largest

Answer

Training a robot to win a board game

Question 2

Clustering belongs to which category of machine learning?

Accepted Answer

Unsupervised learning

Answer

Supervised learning

Answer

Reinforcement learning

Answer

Transfer learning

Question 3

What is a centroid in K-Means clustering?

Accepted Answer

The average center position of all points in a cluster

Answer

The data point farthest from all others in the dataset

Answer

The first data point added to a cluster

Answer

A label assigned to each data point by a human

Question 4

What does the 'K' in K-Means represent?

Accepted Answer

The number of clusters the algorithm should find

Answer

The total number of data points in the dataset

Answer

The distance between the two farthest data points

Answer

The number of features (columns) in the data

Question 5

What is the very first step of the K-Means algorithm?

Accepted Answer

Randomly place K centroids in the data space

Answer

Calculate the silhouette score to evaluate the data

Answer

Sort all data points from largest to smallest

Answer

Remove all outliers from the dataset

Question 6

How is the similarity between two data points measured in most clustering algorithms?

Accepted Answer

Using Euclidean distance, the straight-line gap between them

Answer

By comparing pre-assigned category labels

Answer

By counting how many identical feature values they share

Answer

By sorting both points alphabetically by their names

Question 7

What two things happen during each iteration (round) of K-Means?

Accepted Answer

Points are assigned to the nearest centroid, then centroids move to the mean of their cluster

Answer

New random data points are added to the dataset each round

Answer

K is increased by 1 every round until all points stand alone

Answer

The algorithm deletes the smallest cluster in every round

Question 8

How does K-Means know when to stop running?

Accepted Answer

When the centroids stop moving and cluster assignments stabilize

Answer

When all data points have the same numerical value

Answer

When exactly half the data points have been processed

Answer

When the user manually presses a stop button

Question 9

What is the Elbow Method and what problem does it solve?

Accepted Answer

Choosing the best K by looking for a bend in the inertia-vs-K graph

Answer

Finding the data point that is farthest from all centroids

Answer

Removing outliers before clustering begins

Answer

Measuring how quickly the algorithm converges

Question 10

Why is it important to scale (normalize) your features before running K-Means?

Accepted Answer

Without scaling, large-valued features unfairly dominate the distance calculation

Answer

Scaling makes the dataset smaller so it loads faster

Answer

Scaling is only needed when the dataset has missing values

Answer

Scaling automatically converts text columns into numbers

Question 11

What is an outlier and how can it hurt K-Means clustering?

Accepted Answer

A data point far from all others that can skew centroid positions

Answer

A data point located exactly at its cluster centroid

Answer

An extra feature engineered to improve cluster separation

Answer

The last data point processed in a given iteration

Question 12

What is DBSCAN and how does it differ from K-Means?

Accepted Answer

A density-based method that finds clusters of any shape and marks isolated points as noise, without needing K upfront

Answer

A supervised classification algorithm that needs labeled training data

Answer

An algorithm guaranteed to produce exactly two clusters

Answer

A sorting algorithm that ranks data points by popularity

Question 13

What is hierarchical clustering?

Accepted Answer

A technique that builds a tree of nested clusters by merging or splitting, visualized as a dendrogram

Answer

A method that works only on numerical data, never on text

Answer

An algorithm that always requires you to set K before training

Answer

A method for sorting data alphabetically in a tree structure

Question 14

What is a dendrogram and how do you use it?

Accepted Answer

A tree diagram showing step-by-step cluster merges, where you cut horizontally to choose the number of clusters

Answer

A bar chart showing how many points are in each cluster

Answer

A colored scatter plot used to evaluate K-Means results

Answer

A confusion matrix used to measure classification accuracy

Question 15

What does the silhouette score measure in clustering?

Accepted Answer

How well each point fits its own cluster vs neighboring clusters, scored from -1 to +1

Answer

The total number of iterations K-Means needed to converge

Answer

The straight-line distance between the two farthest centroids

Answer

The percentage of data points classified as outliers

Question 16

What is 'inertia' in K-Means and why does it matter?

Accepted Answer

The sum of squared distances from each point to its nearest centroid

Answer

The speed at which K-Means centroids converge each run

Answer

The number of times centroids move during training

Answer

The probability that a data point belongs to a given cluster

Question 17

Why does K-Means struggle with clusters that are not round (non-spherical)?

Accepted Answer

K-Means assumes clusters are round blobs, so elongated or irregular shapes get split or merged incorrectly

Answer

K-Means cannot process datasets with more than 100 data points

Answer

K-Means is too slow to work on any data that is not numerical

Answer

K-Means needs labeled data in order to handle non-spherical clusters

Question 18

How can clustering be used to compress an image?

Accepted Answer

K-Means replaces every pixel's color with the nearest cluster-centroid color, reducing the total number of colors used

Answer

Clustering deletes all dark pixels to make the file smaller

Answer

Clustering splits the image into equal-sized rectangular blocks

Answer

Clustering converts a color image automatically to black and white

Question 19

Which real-world scenario is the most classic example of clustering?

Accepted Answer

Grouping customers by shopping behavior to enable targeted marketing

Answer

Predicting whether a new email is spam or not

Answer

Teaching a self-driving car to recognize stop signs

Answer

Translating English sentences into French automatically

Question 20

In simple terms, what causes two data points to end up in the same cluster?

Accepted Answer

They are closer to each other than to points in other clusters

Answer

They were both added to the dataset at the same time

Answer

They share the same human-assigned label in the training data

Answer

They both appear in the first column of the data table

Question 21

What is Manhattan distance and how does it differ from Euclidean distance?

Accepted Answer

The grid-path distance calculated as the sum of absolute differences along each dimension

Answer

The straight-line diagonal distance between two points

Answer

The number of matching feature values between two data points

Answer

The distance measured only along the vertical axis

Question 22

Why can running K-Means twice on the same data give different results?

Accepted Answer

Random initial centroid placement can lead the algorithm to different local solutions each run

Answer

The dataset values change randomly between runs

Answer

K is automatically increased by 1 in each new run

Answer

The distance metric is randomly swapped between Euclidean and Manhattan each run

Question 23

What improvement does K-Means++ make over standard random initialization?

Accepted Answer

K-Means++ spreads initial centroids far apart using distance-weighted random selection to avoid bad starts

Answer

K-Means++ automatically selects the best K value without any user input

Answer

K-Means++ uses Manhattan distance instead of Euclidean distance throughout

Answer

K-Means++ adds synthetic data points to balance under-represented clusters

Question 24

In DBSCAN, what is a 'core point' and why does it matter?

Accepted Answer

A point that has at least minPts neighbors within radius ε, forming the foundation of a dense cluster

Answer

The centroid of a cluster, calculated as the average position of all its points

Answer

The data point with the highest numerical value in the entire dataset

Answer

The very first data point selected when the DBSCAN algorithm begins

Question 25

How is clustering fundamentally different from classification?

Accepted Answer

Classification predicts known categories from labeled examples; clustering discovers hidden groups without labels

Answer

Clustering uses labeled data while classification does not

Answer

Both clustering and classification always require labeled training data

Answer

Classification groups data by distance; clustering predicts exact numerical values

Question 26

How do music streaming apps use clustering to recommend songs?

Accepted Answer

Songs are grouped by similar audio features, and listeners get recommendations from the same cluster

Answer

Clustering is used to delete songs that fall below a popularity threshold

Answer

Clustering determines how many advertisements to play between songs

Answer

Songs are clustered so they download in alphabetical order

Question 27

How do biologists use clustering in genetics research?

Accepted Answer

Genes with similar behavior across conditions are grouped together, revealing possible shared biological functions

Answer

Clustering sorts DNA samples by how old they are

Answer

Clustering directly predicts whether a patient will recover from a disease

Answer

Biologists use clustering to label every cell with its exact gene count

Question 28

What is an important ethical concern when clustering is applied to groups of people?

Accepted Answer

Grouping people by data can reinforce bias and cause unfair treatment if cluster features proxy for protected characteristics

Answer

Clustering people is unethical only when the dataset has missing values

Answer

The main concern is that clustering runs too slowly on large human datasets

Answer

Clustering is always fair because it uses only mathematical distances, not human judgment

Question 29

What happens to inertia as you increase K beyond the ideal number of clusters?

Accepted Answer

Inertia keeps decreasing but with diminishing gains, making extra clusters less meaningful

Answer

Inertia increases sharply once K exceeds the ideal number

Answer

Inertia stays perfectly constant once the correct K is found

Answer

Inertia fluctuates randomly as K increases beyond the optimal value

Question 30

What problem arises when two clusters overlap significantly, and how can it be addressed?

Accepted Answer

Points near the boundary are ambiguously assigned; soft clustering methods solve this with probability memberships

Answer

The algorithm runs faster because fewer distance calculations are needed near borders

Answer

All overlapping points are automatically deleted as outliers

Answer

The two overlapping centroids automatically merge into a single centroid

Test yourself