en ‘This question paper contains 7 printed pages) Roll No. §. No. of Question Paper = 2780 ‘Unique Paper Code : 32347611 ic Name of the Paper + Data Mining, ‘Name of the Course : B.Sc. (H) Computer Science : DSE-4 Semester ak ‘Duration : 3 Hours Maximum Marks : 75 (write your Roll No. on the z snmediately on receipt of this question PAPE J Attempt All questions from Section A. ‘Attempt any four questions from Section B. Section A 1 0 6@~«~Find the Euclidean distance between data points x, -1, 0, 1) and YO, 0, -} 0). 2 (b) Wrecall and precision are 0.5 and 0.6 respectively, compute the value of F, measure: 2 ( ma given dataset, it is found that an jtemset {ab} is infrequent. Will itemset {abe} be infrequent or frequent ? Explain why- 2 PTO.@ oO (2) @), @ (2) 2780 What are the three strategies for handling missing values jn a dataset ? E 3 pifferentiate petween precision and bias 0 the basis of the quality of the measurement process: a what is meant by variable transformation 9 What are its advantages 2 3 Af support of an association rule X 7 ¥ is 80% and confidence js 75%, cam we derive support and confidence of the rule yok? if yes, list down the values. If nO, state the reason. 3 List down wo advantages and nwo disadvantages of leave- one-out approach used in cross-validation for evaluating the performance of the classifier 7 4 Differentiate petween agglomerative and divisive methods of hierarchical clustering with the help of a diagram. 4 What are asymmetric. attributes 9 Give an example of each + 4 wo asymmetric binary attribute, Gi) asymmetric diserete attribute, (ii) asymmetric continuous attribute.(®) () © (3) 2780 The confusion matrix for a 2-class problem is given 5 Calculate the Accuracy, Sensitivity, Specificity, True Positive Rate, and False Positive rate. Section B What are the differences between noise and outliers ? ‘Are noise objects always outliers ? Are outliers always noise objects ? +141 Let A and B be two sets of integers. A distance measure ‘d@ is defined as follows = 4 (A —B) = size (A — B) + size (B— A) where ‘-” denotes set difference. Size denotes the number of elements in the set. Prove that the distance measure ‘d is a metric. ‘What is unsupervised learning ? Explain with the help of an example application. 2 PTO. ienseeiiemmens: “6 — 7(4) 2780 4 @ Consider the following dataset for a 2-class problem : 7 qi) Gio) Calculate the gain in the Gini Index when Splitting on A and B. Which attribute would the decision tree induction algorithm choose 2 Draw the decision tree after splitting showing the number of instances of each class.CC (6) 4. @ ® © 5. (@) (5) 2780 (iv) How many instances are misclassified by the resulting decision tree ? Why is K-nearest neighbor classifier a lazy learner ? 3 What is an exhaustive rule-sets in Rule based classi- fication ? If the rule-set is not exhaustive, what problem arises ? How is it resolved 7 4 What is progressive sampling ? What are its advantages ? 7 State Bayes" theorem. What assumption is used by the Naive Bayes classifier ? 3 Consider the following set of frequent 3-itemsets : (1, 2, 3}, 41, 2, 4}, {1, 2, 5} 41, 3,4}, €1, 3, 5}, (2, 3,43, 42.3, 5}, 43, 4, 5}. Assume that there are only five items in the dataset. () List all candidate 4-itemsets obtained by a candidate generation procedure using the F,_, x F, merging strategy. (ii) Listall candidate iter ‘obtained by a candidate generation procedure in Apriori. 6 eS(6) 2780 (6) Let X denotes the categorical attribute having values {awful, poor, OK, good}. What is the representation of each value when X is converted to binary form using : @ 2 bits (i) 4 bits 2 = 4 6 Consider the following transactional dataset = 8 Transaction ID Items Bought 0001 fa, d, eh 0002 {a, 5, c, e} 0003 ta, b, a, ef 0004 fa, ¢, d, e} 0005 {b,c e} 0006 fb, d, e} 0007 fe, d} 0008, {a, b, e} 0009 {a, d, e} 0010 : {a, b, e}(b) @ (6) a (7) 2780 vw) Find out the support of itemsets fey, {5s qd, {a, dy and {b, d, e}. Are these itemsets frequent if minimum support threshold is 30% ? i) Find all the rules. generated from the 3-itemset {b. a, eh List down the strong rules among these rules if minimum confidence threshold is 60%. What is the difference petween nominal attributes and ordinal attributes 9 Give an example of each. iz Explain the following terms with reference t0 the DBSCAN clustering. algorithm * (@ Core point (i) Noise point (iid) Border point 6 Given the following dat2 points : 2, 4, 10, 12, 3, 20, 30, 11,25. ‘Assume K = 3 and initial means 2, 4, 6. Show the clusters obtained using K-means algorithm after two iterations aid show the new means for the next iteration. 4