Font size
WorksheetsAI502_ML_TA_1_AIDS
Total questions: 30
Worksheet time: 20mins
Which of the following is NOT a goal of Machine Learning?
Automate decision-making processes
Learn from data
Write rules manually
Improve performance with experience
Why is Machine Learning gaining popularity?
Data is decreasing rapidly
It avoids use of mathematics
High computation power and availability of big data
It replaces all software engineers
Which is a major reason for using ML over traditional programming?
Less accuracy
Data-driven learning
Manual rule creation
Predefined outputs
In Machine Learning, what is a ‘model’?
A physical machine
An algorithm trained on data to make predictions
A mathematical puzzle
· A design template
The process of training a model involves:
Running code without data
Feeding data to the model and optimizing its performance
Drawing charts manually
Building a database
Which is an example of supervised learning?
Clustering customers
Stock price prediction using historical data
Recommending movies without labeled data
Self-driving car learning through exploration
Write an formula to obtained accuracy in binary classification problem of ML
Which of the following is an example of reinforcement learning?
Image classification
Grouping similar documents
A robot learning to walk by trial and error
Predicting exam scores
Which of the following can be considered as a quality issue in data?
Complete data
Redundant features
Clean data
Standardized formats
What is the main goal of data preprocessing?
Increase model complexity
Prepare raw data for modeling
Reduce training time only
Add random noise to data
Which of the following is a feature scaling technique that brings all values between 0 and 1?
Standardization
Label Encoding
Min-Max Normalization
Ordinal Encoding
(a) encoding assigns unique numbers to each category without considering any order.
The formula for Z-score standardization is (x - mean) / __________.
Standard Deviation
Mode
Median
Z/x
(a) feature selection methods evaluate features based on statistical tests before modeling.
Which of the following is NOT a type of Missing Data Technique
Deletion
Missing Completely at Random (MCAR)
Mean Imputation
Missing at Random (MAR)
Missing Not at Random (MNAR)
You are given a dataset with missing values in the "Age" column. Which method would you use if the data is normally distributed?
Mode imputation
Mean imputation
Median imputation
Remove the column
What is the main drawback of One-Hot Encoding?
Loss of information
Cannot handle numerical data
It introduces high dimensionality
It requires missing data
Z-score normalization assumes that data is:
Uniformly distributed
Normally distributed
Positively skewed
Random
In K‑NN classification, as K increases, what happens to bias and variance?
Bias increases, variance decreases
Bias decreases, variance increases
Both increase
Both decrease
A decision tree reaches 100% training accuracy. Which of the following models trained on the same data will also achieve 100% accuracy?
KNN
SVM Polynomial-kernel
Naive Bayes
None of the above
Which of the following best describes bagging?
Sequential boosting of weak learners
Averaging predictions over multiple bootstrap samples
Stacking heterogeneous base models
Feature selection technique
Which metric is not derived from the confusion matrix in classification?
Precision
Recall
ROC-AUC
Error Rate
High bias, low variance is a scenario corresponds to (a) in terms of bias and variance
Which attribute will a decision tree select first when building the tree using the ID3 algorithm?
The one with the lowest entropy
The one with the highest information gain
The one that occurs most frequently
The one with the lowest mean value
Which of the following best describes entropy in information theory?
It measures the distance between clusters
It is the measure of randomness or impurity in the dataset
It is used to calculate the mean of data
It is used to evaluate the accuracy of models
What is the main objective of a Support Vector Machine (SVM)?
To minimize the number of support vectors
To find the hyperplane that maximizes the margin between classes
To maximize the error in classification
To reduce the number of features
In the context of SVM, what are support vectors?
Vectors that support the training process
Data points that are farthest from the hyperplane
Data points that lie closest to the hyperplane and influence its position
Data points that are always misclassified
Which of the following kernels is not commonly used in SVM?
Linear kernel
Polynomial kernel
Radial Basis Function (RBF) kernel
Exponential Moving Average kernel
In K-Fold Cross Validation, if k = 5, how many models are trained?
1
2
5
10
In Decision Trees, which strategy helps reduce overfitting?
Expanding the tree until all leaves are pure
Using all available features
Pruning the tree
Increasing the depth of the tree
