Data Mining and Machine Learning
CSE · Final · Spring 25
Batch 62
Read the questions as text
Read off the PDF by AI, so it may have mistakes. The PDF is the original.
Daffodil International University
Faculty of Science & Information Technology
Department of Computer Science & Engineering
Final Semester Examination, Spring 2025
Course Code: CSE325, Course Title: Data Mining and Machine Learning
Level: 3 Term: 2 Batch: 62
Time: 2:00 Hrs
Marks: 40
Answer ALL Questions
[The figures in the right margin indicate the full marks and corresponding course outcomes. All portions of each question must be answered sequentially.]
1. The following dataset contains shopping transactions from an online fashion store. Your task is to apply the FP-Growth algorithm to extract all frequent itemsets that satisfy a minimum support threshold of 50%. | CO1
Transaction ID | Items Purchased
T1 | T-Shirt, Sneakers, Jeans, Cap, Sunglasses
T2 | Jacket, Sneakers, T-Shirt, Jeans
T3 | Sunglasses, Watch, Sneakers, Jeans
T4 | T-Shirt, Sneakers, Belt, Cap
T5 | Jeans, Jacket, Watch, Sneakers, T-Shirt
T6 | T-Shirt, Sneakers, Watch, Sunglasses
T7 | Cap, Sneakers, Jeans, T-Shirt, Jacket
a) Construct the FP-Tree based on item frequency and item ordering. | [5]
b) Explain how conditional pattern bases and conditional FP-Trees are formed. | [3]
c) Discuss the efficiency of FP-Growth over Apriori, especially for dense datasets. | [2]
2. You are given a set of 15 two-dimensional points representing spatial coordinates of tracked devices in a smart campus. Apply the DBSCAN algorithm using the following parameters: minPts = 4, eps (e) = 2.0. | CO1
P1(1,6), P2(2,5), P3(3,7), P4(4,6), P5(5,5), P6(5,3), P7(4,2), P8(6,2), P9(2,2), P10(1,3), P11(3,3), P12(6,5), P13(7,4), P14(8,5), P15(7,6)
a) Identify all core points and border points based on the parameters provided. | [2.5]
b) Use the DBSCAN algorithm to form clusters - i) List all clusters and ii) Indicate which points are noise (if any) | [2.5]
3. You are given the following dataset consisting of 10 samples with 3 features: | CO1
Sample | Feature A | Feature B | Feature C
A | 2 | 0 | 1
B | 0 | 1 | 4
C | 4 | 3 | 2
D | 3 | 5 | 3
E | 1 | 2 | 2
F | 5 | 1 | 0
G | 2 | 4 | 1
H | 3 | 2 | 4
I | 1 | 0 | 3
J | 0 | 3 | 2
a) Standardize the dataset so that each feature has zero mean and unit variance. | [2.5]
b) Compute the covariance matrix of the standardized dataset. | [2.5]
c) Find the eigenvalues and eigenvectors of the covariance matrix. | [2.5]
d) Identify the principal components in descending order of variance explained. | [2.5]
4. An artificial neural network has been trained on a binary classification dataset where the independent attributes are Feature A and Feature B. The target classes are encoded as Class A = 0 and Class B = 1 before training. In this network: (i) All hidden nodes use the ReLU activation function, and (ii) The output node uses the Sigmoid activation function
The trained network is shown below. Classify the data samples in Table 1 using this network. | CO2
Table 1: Test Data
Sample | Feature A | Feature B
S1 | 1.2 | 2.5
S2 | 2.0 | 1.0
Neural Network Architecture
- Input Layer: Feature A, Feature B, and Bias
- Hidden Layer: 3 neurons (H1, H2, H3)
- Output Layer: 1 neuron (O)
Input to Hidden Weights:
From -> To | H1 | H2 | H3
Feature A | -1.5 | 1.0 | 2.1
Feature B | 2.3 | -0.7 | 1.8
Bias | -1.0 | -1.2 | -0.5
Hidden to Output Weights:
From -> To | Output
H1 | 2.0
H2 | -1.5
H3 | 1.3
Bias | -2.0
a) Compute the input to each hidden neuron using the given weights and bias. | [2.5]
b) Apply the ReLU activation to hidden neurons. | [2.5]
c) Compute the input to the output neuron and apply the Sigmoid activation. | [5]
d) Classify the result as Class A (0) or Class B (1) based on threshold 0.5. | [5]