Data Mining and Machine Learning
CSE · Final · Fall 24
Batch 61
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, Fall 2024
Course Code: CSE325, Course Title: Data Mining & Machine Learning
Level: 3 Term: 2 Batch: 61
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. a) Analyze the following transaction dataset to derive all frequent itemsets with a minimum support & Confidence threshold of 50%. Justify your approach by evaluating the intermediate steps and explaining the computational challenges encountered. | [Marks: 10] | CO3
Transaction ID | Items Purchased
T1 | Bread, Butter, Milk, Eggs
T2 | Bread, Butter, Jam
T3 | Milk, Bread, Butter, Eggs
T4 | Butter, Milk, Eggs
T5 | Bread, Butter, Eggs, Jam
T6 | Bread, Milk, Eggs
T7 | Bread, Butter, Milk, Jam
T8 | Butter, Jam
T9 | Bread, Butter, Milk, Eggs
T10 | Butter, Milk, Jam
2. a) Explain the concept of cross-validation and discuss how it contributes to improving a model's generalization ability on new data. | [Marks: 3] | CO2
b) Demonstrate the challenges of using regular K-Fold Cross-Validation with a highly imbalanced dataset. Show how Stratified K-Fold Cross-Validation can be applied to resolve these issues effectively. | [Marks: 7]
3. a) Define Hierarchical Agglomerative Clustering (HAC) and identify any two methods used to measure the distance between clusters in HAC. | [Marks: 2] | CO3
b) Construct a Dendrogram for the dataset below using HAC with single-linkage. Data-points are A1(1,3), A2(2,4), A3(3,5), A4(7,8), A5(8,9), B1(1,2), B2(4,6). | [Marks: 8]
4. a) Analyze Forward Propagation on the following Neural Network given below and calculate the loss. The actual output of y8 = 0.01 and y9 = 0.99 and the neurons are using sigmoid activation function. | [Marks: 10] | CO4
[Figure: A neural network with 3 input nodes (X1=0.05, X2=0.10, X3=0.20), two hidden layers with hidden nodes H4, H5, H6, H7, bias b1=0.35, various edge weights (w14=0.15, w15=0.25, w34=0.30, w25=0.20, w46=0.40, w56=0.45, w56=0.55, w47=0.50, w56=0.55, w71=0.70, w62=0.60), and two output nodes y8=0.01 and y9=0.99]