Explain the bias–variance trade-off in machine learning. Provide examples of how increasing model complexity affects both bias and variance. [2+3]
Describe the decision tree algorithm. What are the advantages and disadvantages of using decision trees? [3+2]
Discuss the structure and learning process of a neural network in supervised learning.
The ages of 10 employees in a company are: [2.5+2.5]
22, 25, 29, 30, 31, 33, 35, 36, 40, 45
a. Find minimum, Q1, median, Q3, and maximum values.
b. Draw a box plot for the data.
A spam filter uses the words "offer" and "win" as features.
From training data:
| Class | P(offer) | P(win) | Prior Probability |
| Spam | 0.8 | 0.6 | 0.4 |
| Not Spam | 0.1 | 0.05 | 0.6 |
Given an email containing both words, classify it as Spam or Not Spam using Naïve Bayes.
Explain the steps of the Principal Component Analysis (PCA) algorithm.
Perform one iteration of K-means clustering for the following points with initial centroids C1 = (1,1) and C2 = (5,4): Points: (1,1), (2,1), (4,3), (5,4). Use Euclidean distance.
Attempt any TWO questions
[2x10=20]Describe the Support Vector Machine (SVM) concept of maximum margin classification with a diagram. A model trained on a dataset achieves the following results:
Training accuracy: 95%
Validation accuracy: 80%
Test accuracy: 78%
Discuss whether the model is underfitting, overfitting, or well-fitted. Justify your answer. [6+4]
Explain the concept of Confusion Matrix with example. Given the dataset below, fit a linear regression model using the least squares method and find the best-fit line equation. Also find the value of Y at X=4.5 [3+7]
| X | Y |
| 1 | 2 |
| 2 | 4 |
| 3 | 5 |
| 4 | 6 |
| 5 | 8 |
Explain the K-nearest neighbor (KNN) algorithm. How does the value of k affect bias and variance?
Given two classes of points:
Class +1: (2, 2), (4, 4)
Class −1: (4, 0), (0, 0)
Determine the equation of the separating hyper-plane. [6+4]