wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

BUSA3020 Quiz W7-W12

Total questions: 93

Worksheet time: 47mins

Name
Class
Date
1.

What are pipelines in scikit-learn used for?

a)

None of the listed options

b)

Combining multiple classifiers models into one

c)

Combining multiple processing steps into one

d)

Visualising data transformation steps

e)

Generating synthetic data samples

2.

Which of the following is a primary advantage of using pipelines in scikit-learn?

a)

Automatic data collection

b)

Eliminating the need for data validation and testing

c)

None of the listed options

d)

Simplicity and Convenience

e)

Reducing the need for data cleaning

3.

How do pipelines improve the reproducibility of machine learning workflows?

a)

By increasing the accuracy of predictions

b)

By reducing the complexity of algorithms

c)

None of the listed options

d)

By automatically updating data sources

e)

By making preprocessing steps explicit

4.

What is the role of the fit method in a scikit-learn pipeline?

a)

To predict new data points

b)

To visualise the data transformation process

c)

To split the data into training and testing sets

d)

To pass data through transformers and train the estimator

e)

None of the listed options

5.

What does the predict method in a scikit-learn pipeline do?

a)

Determines the best hyperparameters for the model

b)

None of the listed options

c)

Calculates the accuracy of the model

d)

Do intermediate steps and predict using the final estimator

e)

Fits the data to the pipeline

6.

What does the holdout method in machine learning involve?

a)

Only using a training set to evaluate the model

b)

Using the entire dataset for both training and testing

c)

Evaluating the model using cross-validation only

d)

Split of the dataset into training, validation, and test sets

e)

None of the listed options

7.

K-fold cross-validation improves upon the holdout method by

a)

Using each fold as a validation set exactly once

b)

Only using one fold for both training and testing

c)

None of the listed options

d)

Eliminating the need for a validation set

e)

Reducing the computational complexity of model evaluation

8.

The need for cross-validation arises due to

a)

The high cost of data collection

b)

The simplicity of machine learning models

c)

None of the listed options

d)

The availability of large datasets

e)

Limitations of the holdout method

9.

What is the purpose of hyperparameter tuning via grid search?

a)

To reduce the size of the dataset

b)

To increase the speed of the training process

c)

To select the best machine learning algorithm

d)

None of the listed options

e)

To find the best combination of hyperparameters values

10.

What do learning curves plot?

a)

Model accuracy versus the sample size

b)

Model accuracy versus the number of features

c)

The number of iterations versus the error rate

d)

The learning rate versus the validation accuracy

e)

None of the listed options

11.

What can validation curves help identify in a model?

a)

None of the listed options

b)

The best machine learning algorithm

c)

The most important features

d)

The optimal size of the dataset

e)

Overfitting and underfitting

12.

Which of the following is a component of a confusion matrix?

a)

True Positive Rate

b)

None of the listed options

c)

False Negative Rate

d)

Unlabeled samples

e)

False Discoveries

13.

Precision in the context of classification models is:

a)

The ratio of true positives observations to the total predicted positives

b)

The ratio of correctly predicted positive observations to the total observations

c)

The ratio of correctly predicted negatives to the total predicted negatives

d)

None of the listed options

e)

The accuracy of the model

14.

The F1 score is used to

a)

Determine the learning rate

b)

Calculate the model's training speed

c)

Balance precision and recall

d)

None of the listed options

e)

Measure the model's accuracy only

15.

Recall is calculated as

a)

TP/(TP+FP)

b)

TN/(TN+FP)

c)

None of the listed options

d)

TN/(TN+FN)

e)

TP/(FN+TP)

16.

Which of the following best defines a centroid in clustering?

a)

The average of all points in a cluster

b)

The most distant point in a cluster

c)

The point with maximum variance

d)

None of the listed options

e)

The least central point in a cluster

17.

A medoid is best defined as

a)

Always positioned at the edge of a cluster

b)

A randomly chosen point in a cluster

c)

None of the listed options

d)

An average point in multidimensional space

e)

The most centrally located point within a cluster

18.

What is the main advantage of K-Means++ over K-Means?

a)

It spreads out the initial centroids

b)

It selects initial centroids randomly

c)

It produces more clusters

d)

It requires fewer clusters

e)

None of the listed options

19.

The K-Means algorithm is not suited for

a)

Assigning data points to the nearest cluster

b)

Finding the longest distance between clusters

c)

None of the listed options

d)

Partitioning n items into k clusters

e)

Minimising within-cluster variances

20.

Euclidean distance is used to

a)

Measure the total variance within a cluster

b)

Calculate the density of clusters

c)

Determine the highest point in a dataset

d)

None of the listed options

e)

Quantify the dissimilarity between points in space

21.

The Elbow method is used to determine

a)

The best distance metric for clustering

b)

The optimal number of clusters in a dataset

c)

The maximum number of iterations for K-Means

d)

The number of outliers in a dataset

e)

None of the listed options

22.

Silhouette plots examine

a)

None of the listed options

b)

The number of iterations needed for convergence

c)

The similarity of features within a cluster

d)

The quality of clustering

e)

The maximum distance between clusters

23.

Hierarchical clustering does not include

a)

Organising clusters as trees

b)

Random assignment of clusters

c)

None of the listed options

d)

Agglomerative (Bottom-Up) Method

e)

Divisive (Top-Down) Method

24.

The complete linkage method

a)

Defines the distance between two clusters as the maximum distance between any single data point in the first cluster and any single data point in the second cluster

b)

Calculates the distance between clusters based on the shortest distance from any member of one cluster to any member of the other cluster

c)

Merges clusters based on the average distance between all pairs of points in the two clusters

d)

Is a type of agglomerative clustering

e)

Defines the distance between two clusters as the distance between their centroids

25.

A dendrogram is used in

a)

None of the listed options

b)

Only divisive clustering

c)

Only K-Means clustering

d)

Illustrating the arrangement of hierarchical clustering

e)

Displaying the exact coordinates of data points

26.

DBSCAN algorithm classifies clusters based on

a)

The number of centroids

b)

None of the listed options

c)

The density of data points

d)

The similarity of feature-target variable pairs

e)

The maximum distance between any two points

27.

Which of the following is not a key concept of DBSCAN?

a)

None of the listed options

b)

Noise Point

c)

Core Point

d)

Border Point

e)

Centroid Point

28.

What does the initial step in K-Means clustering involve?

a)

Evaluating the model's accuracy

b)

Selecting k centroids

c)

Calculating the cluster count

d)

Assigning data points to existing clusters

e)

None of the listed options

29.

Which of the following is not a goal of unsupervised learning?

a)

To predict the outcome for new, unseen data

b)

To find clusters or associations without knowing the true labels

c)

To explore the structure of the data to find patterns

d)

None of the listed options

e)

To model the underlying structure or distribution in the data

30.

In the context of Euclidean distance, what is d(y,x) if y and x are vectors?

a)

The maximum distance between two points in a dataset

b)

The product of two vectors y and x

c)

The average position of all points in a cluster

d)

None of the listed options

e)

The length of the line segment connecting x and y

31.

What distinguishes the Elbow method in cluster analysis?

a)

None of the listed options

b)

It identifies the point where increasing the number of clusters significantly improves the fit of the data

c)

It calculates the number of points in each cluster

d)

It determines the minimum Euclidean distance for optimal clustering

e)

It identifies the point where increasing the number of clusters does not significantly improve the fitting of the data

32.

Which statement about DBSCAN is false?

a)

It requires pre-specifying the number of clusters

b)

None of the listed options

c)

It classifies points as core points, border points, or noise points

d)

Core points have a minimum number of points within a given radius

e)

It is based on the density of data points

33.

What does an ensemble method involve in machine learning?

a)

Focusing solely on reducing variance in predictions

b)

Ensemble methods are designed to

c)

Using only different types of models within model sets

d)

Generating multiple models and combining their predictions

e)

None of the listed options

34.

Ensemble methods are designed to

a)

Reduce only bias

b)

None of the listed options

c)

Reduce only variance

d)

Increase both bias and variance

e)

Focus solely on model accuracy

35.

What is majority voting in the context of ensemble methods?

a)

When more than half of the classifiers agree on the same class

b)

When each classifier has one vote, regardless of confidence

c)

None of the listed options

d)

When only one classifier decides the final prediction

e)

When a class needs 100% of the votes to be chosen

36.

Plurality voting is used when

a)

There's only one possible outcome

b)

The class with more than 50% of the wins

c)

None of the listed options

d)

The class with the most votes wins

e)

A class receives exactly half of the votes

37.

Hard voting in predictive ensembles is characterised by

a)

Ignoring the votes from less accurate classifiers

b)

Considering only the classifiers' confidence levels

c)

Summing the weighted votes for each class label

d)

None of the listed options

e)

Averaging probability estimates from classifiers

38.

Soft voting differs from hard voting by

a)

Requiring unanimous agreement among classifiers

b)

Using only the most confident classifier's prediction

c)

None of the listed options

d)

Averaging the probability estimates for each class label

e)

Not using votes from classifiers

39.

Bagging is an ensemble method that

a)

Can reduce bias without significantly increasing variance

b)

None of the listed options

c)

Increases variance to avoid overfitting

d)

Can reduce variance without significantly increasing bias

e)

Trains each model on the entire dataset without replacement

40.

A weak learner is a model that

a)

Performs worse than random guessing

b)

None of the listed options

c)

Is always a model with high bias

d)

Is always a model with high variance

e)

Cannot be combined to improve performance

41.

Adaptive Boosting (AdaBoost) works by

a)

Reducing variance while increasing bias

b)

Combining strong learners into a single model

c)

Focusing solely on correctly classified instance

d)

None of the listed options

e)

Combining multiple weak learners to form a strong learner

42.

In AdaBoost, the weights of instances are updated to

a)

None of the listed options

b)

Decrease emphasis on misclassified instances

c)

Give more emphasis to misclassified instances

d)

Ensure all instances have equal weight

e)

Give more emphasis to correctly classified instances

43.

AdaBoost is sensitive to noisy data and outliers because

a)

It focuses on instances that are hard to classify

b)

It assigns lower weights to them automatically

c)

It ignores them during the training process

d)

It uses them to reduce the model's bias

e)

None of the listed options

44.

The main difference between Bagging and AdaBoost is that Bagging

a)

None of the listed options

b)

Is sensitive to noisy data and outliers

c)

Reduces bias only

d)

Creates a set of dependent classifiers

e)

Reduces variance without significantly affecting bias

45.

AdaBoost updates the instance weights by

a)

Randomly assigning weights in each iteration

b)

Keeping the weights constant throughout the learning process

c)

None of the listed options

d)

Increasing weights for misclassified instances

e)

Decreasing weights for instances hard to classify

46.

What is the role of weights in soft voting?

a)

They have no role in soft voting

b)

They are used to average the probability estimates for class labels

c)

They decide which classifiers to exclude from the ensemble

d)

None of the listed options

e)

They determine the final prediction through majority voting

47.

Which statement is not true about majority voting in ensemble methods?

a)

It applies to binary classification problems

b)

It always leads to correct classification

c)

It is a method to aggregate predictions from multiple models

d)

None of the listed options

e)

It requires more than 50% of the votes for a decision in binary classification

48.

Which of the following is not true about soft voting?

a)

It averages probability estimates for each class label

b)

It relies on the confidence levels of classifiers' predictions

c)

It considers only the class labels predicted by each classifier

d)

None of the listed options

e)

It can involve weighted averaging of probabilities

49.

Which statement is NOT true regarding AdaBoost?

a)

None of the listed options

b)

It can be sensitive to noise and outliers

c)

It involves an iterative reweighting of instances

d)

It focuses more on instances that are harder to classify

e)

It combines multiple strong learners to reduce bias

50.

Which of the following is not true?

a)

Simple regression requires more data

b)

Multiple regression is used when there are no confounding variables

c)

All of the listed options

d)

Simple regression is used for predicting multiple outputs

e)

Multiple regression is computationally less intensive

51.

What of the following techniques is used in Exploratory Data Analysis (EDA)?

a)

Inferential statistics

b)

Descriptive statistics

c)

Predictive modelling

d)

None of the listed options

e)

Validation analysis

52.

What do scatter plots primarily show?

a)

The relationship between two variables

b)

The spread/variance of a univariate variable

c)

None of the listed options

d)

The central tendency of a univariate variable

e)

The correlation coefficient between two variables

53.

What does a correlation coefficient of zero indicate?

a)

Perfect negative linear relationship

b)

No relationship

c)

Perfect positive linear relationship

d)

No linear relationship

e)

High degree of correlation

54.

How is the correlation matrix useful in multiple regression analysis?

a)

Can aid in building a regression model

b)

Helps detect multicollinearity

c)

Visualising data connections

d)

All of the listed options

e)

Feature selection

55.

Why is the RANSAC algorithm preferred in the presence of data anomalies?

a)

None of the listed options

b)

It uses less data

c)

It reduces computational time

d)

It is robust to outliers

e)

It handles multiple outputs

56.

Which method is not used to evaluate the performance of a linear regression model?

a)

Residual Plots

b)

None of the listed options

c)

Correlations

d)

R-squared (R²)

e)

Mean Squared Error (MSE)

57.

What does Lasso Regression do differently from Ridge Regression?

a)

Uses squared coefficients in the penalty

b)

None of the listed options

c)

Shrinks more coefficients to zero

d)

Uses a combination of L1 and L2 penalties

e)

Is computationally efficient

58.

Which is a potential disadvantage of polynomial regression models?

a)

None of the listed options

b)

Risk of overfitting

c)

Cannot handle interactive terms

d)

Cannot model linear relationships

e)

Too simple for complex datasets

59.

What is an advantage of Random Forest Regression over single Decision Tree Regression?

a)

Random Forests are not linear

b)

Easier to compute

c)

Requires less data

d)

Less likely to overfit

e)

None of the listed options

60.

Which regression method includes both L1 and L2 regularisation techniques?

a)

Quadratic Polynomial Regression

b)

None of the listed options

c)

Elastic Net Regression

d)

Lasso Regression

e)

Ridge Regression

61.

In the context of regression models, what does RANSAC stand for?

a)

Randomised Statistical Consensus

b)

Random Sample Consensus

c)

Regression Analysis Sample Consensus

d)

None of the listed options

e)

Regression Sample Consensus

62.

Which metric is specifically used to assess the goodness of fit in regression models?

a)

Mean Squared Error (MSE)

b)

Mean Absolute Percentage Error (MAPE)

c)

None of the listed options

d)

Root Mean Squared Error (RMSE)

e)

Median Absolute Absolute (MAE)

63.

Which of the following is true about Random Forest and Gradient Boosting ensemble methods?

a)

Both methods can be only used for classification problems

b)

None of the listed options

c)

Random Forest is used for classification whereas Gradient Boosting is used for regression problems

d)

Both methods can be only used for regression problems

e)

Random Forest is used for regression problems whereas Gradient Boosting is used for Classification problems

64.

The results below are obtained for three different models based on the same dataset. Which model exhibits the overfitting problem?

Model-1

MSE based on training data = 17.19

MSE based on testing data = 20.20

Model-2

MSE based on training data = 39.20

MSE based on testing data = 63.50

Model-3

MSE based on training data = 68.3

MSE based on testing data = 69.4

a)

None of the listed models

b)

Model 2

c)

Model 3

d)

Model 1

e)

All of the listed models

65.

Which of the following best describes time series data?

a)

Data collected over time at regular intervals

b)

Data where observations are always independent of each other

c)

Data collected at a single point in time

d)

Data with no temporal component

e)

None of the listed options

66.

Which type of model is appropriate when the current value of a time series depends only on previous values of the series?

a)

Last historical values of the series

b)

None of the listed options

c)

Autoregressive

d)

White Noise

e)

Moving Average

67.

Which of the following best describes white noise?

a)

No temporal dependence

b)

None of the listed options

c)

High autocorrelation at all lags

d)

Non-zero mean

e)

Positive trend

68.

The ACF (Autocorrelation Function) of a white noise process is:

a)

Zero for all non-zero lags

b)

Gradually decaying towards zero

c)

None of the listed options

d)

Zero for all lags

e)

Non-zero for all lags

69.

In a Moving Average (MA) model, the current value of the time series depends on:

a)

Past error terms

b)

Future values of the series

c)

Past values of the series

d)

None of the listed options

e)

Only the most recent observation

70.

Which of the following models combines both AR and MA components?

a)

Autoregressive Moving Average Model

b)

White Noise

c)

Vector Autoregressive Model

d)

None of the listed options

e)

Generalised Autoregressive Model

71.

If the PACF of a time series cuts off after lag 2 while ACF decays gradually, what model is most likely appropriate?

a)

None of the listed options

b)

AR(2)

c)

AR(1)

d)

ARMA(2,1)

e)

MA(2)

72.

Which of the following statements is true for an AR(1) process?

a)

It depends on the previous value and error term

b)

It depends only on the current error term

c)

None of the listed options

d)

It is a white noise process by definition

e)

It depends on the previous and current error terms

73.

A time series model where the current value depends only on current and past error terms is known as:

a)

White Noise

b)

Moving Average (MA) model

c)

Cross-sectional model

d)

Autoregressive (AR) model

e)

None of the listed options

74.

What does the autocorrelation function (ACF) measure in a time series?

a)

The frequency of seasonal patterns

b)

The correlation between lagged values and current values

c)

The relationship between current and future values

d)

The standard deviation of the time series

e)

None of the listed options

75.

In a time series, seasonality refers to:

a)

None of the listed options

b)

A long-term upward or downward movement in the data

c)

A temporary shock to the series

d)

Regular patterns that repeat at fixed intervals

e)

Random fluctuations with no pattern

76.

Which of the following is true about the Moving Average (MA) model?

a)

The ACF cuts off after lag q

b)

Both ACF and PACF cut off after q lags

c)

None of the listed options

d)

The PACF cuts off after lag q

e)

The ACF gradually decays

77.

Which of the following is true about the residuals of a well-fitted ARMA model?

a)

The residuals should behave like white noise

b)

The residuals should have increasing variance

c)

The residuals should exhibit high autocorrelation

d)

None of the listed options

e)

The residuals should show a trend over time

78.

In a well-fitted ARMA model, the ACF and PACF of the residuals should:
None of the listed options

a)

None of the listed options

b)

Show a gradual decay

c)

Show significant autocorrelation

d)

Should cut off to zero after lag 3

e)

Be zero for all non-zero lags

79.

What does the Partial Autocorrelation Function (PACF) measure?

a)

None of the listed options

b)

The correlation between the current value and future values of the time series

c)

The correlation between the current value and all previous values of the time series

d)

The overall trend in the time series data

e)

The correlation between the current value and lagged values, after removing the effects of intermediate lags

80.

What is Sentiment Analysis primarily used for in the context of machine learning?

a)

Evaluating the grammatical accuracy of text

b)

None of the listed options

c)

Translating text from one language to another

d)

Converting speech into text

e)

Predicting the next word from textual data

81.

Which application of Sentiment Analysis was mentioned in the lecture notes?

a)

Analysing email sentiments for marketing purposes

b)

Sorting product reviews into predefined categories

c)

None of the listed options

d)

Distinguishing between positive and negative movie reviews on IMDb

e)

Building a predictor for financial market trends based on news articles

82.

What does the Bag-of-Words model ignore when processing text?

a)

The frequency of words

b)

The uniqueness of words

c)

The grammar and order of words

d)

None of the listed options

e)

All of the listed options

83.

Which is not a core component of the Bag-of-Words model?

a)

Vocabulary Creation

b)

Tokenization

c)

Syntax Parsing

d)

None of the listed options

e)

Vectorization

84.

What does TF-IDF stand for?

a)

Term Formulation-Inverse Document Frequency

b)

Total Frequency-Integrated Document Frequency

c)

None of the listed options

d)

Term Frequency-Inverse Document Frequency

e)

Text Frequency-Inverse Document Frequency

85.

What is an N-gram model used to represent?

a)

A predictive model for text classification on n words

b)

The frequency of n terms across documents

c)

None of the listed options

d)

The importance of n documents

e)

A sequence of n items from text

86.

What are regular expressions used for in Python?

a)

String search and manipulation

b)

Creating secure passwords

c)

Developing web applications

d)

Calculating numerical statistics

e)

None of the listed options

87.

What is a token in NLP?

a)

None of the listed options

b)

A regex expression in Python

c)

A type of punctuation mark

d)

A semantic unit of analysis

e)

A method for text summarization

88.

What does word stemming achieve?

a)

None of the listed options

b)

Reducing words to their root form

c)

Correcting grammatical errors

d)

Encrypting text data

e)

Expanding the vocabulary of a text

89.

What is the main difference between CountVectorizer and HashingVectorizer?

a)

CountVectorizer uses Python regex; HashingVectorizer does not

b)

HashingVectorizer is used only for large text bodies

c)

CountVectorizer maintains a full dictionary; HashingVectorizer does not

d)

CountVectorizer supports n-grams; HashingVectorizer does not

e)

None of the listed options

90.

What is Topic Modeling used for?

a)

Encoding text data

b)

Discovering sentiment in a collection of texts

c)

Discovering abstract themes in texts

d)

Correcting syntactical errors

e)

None of the listed options

91.

Which technique is known as out-of-core learning?

a)

A method that involves fitting the model to all data

b)

A technique used only in neural network training

c)

A method that involves continuous model training

d)

Incrementally fitting a classifier using batches of data

92.

What is primarily reduced by the TF-IDF calculation?

a)

The influence of common words

b)

The need for manual review of documents

c)

None of the listed options

d)

The accuracy of document classification

e)

The size of the text corpus

93.

Which of the following is NOT true about the Bag-of-Words model?

a)

It considers the semantic meaning of words

b)

It maintains word frequency

c)

It breaks text down into a collection of words

d)

It involves vectorization of documents

e)

None of the listed options