Data Mining — DU B.Sc. (Hons) Computer Science Previous Year Question Papers

: B.Sc. (Hons.) Computer science 4. Parts of a question must be answered together. 1. /cz/ Determine the attribute type for each of the following rz) Movie ratings on a scale of 1 to 5 stars /z.zJ Temperature measured in degrees celsius /z.I.I) PostavzIP codes /z-v/ Number of children in a family. ¢/ Find the Euclidean distancebetweenthe datapoints x(1, 2, 0, -I) and Y(03 -I, 2, 1). rc/ If the recall and precision of a classifier are o.6 and o.8 respectively, compute the value of Fi measure. re/ ct) Briefly describe two methods used for handling missing values in a dataset. How is an eager leaner classifler different from a lazy leaner classifier? Support your answer with one example of each, © Consider a categorical attribute Quality with values {very poor, poor, average, good, (3) excellent} . Show the representation of eacb value when this attribute is converted to birmry form using : /i.jj 5 asymmetric binary attributes. „ Explain the following terms with reference to the DBSCAN cluster.ng algorithm : /z) Corepoint rz.z) Border point /I) Ngrmalize the following s?t ofvalues using min-max normalization : {15,1.9, 22] 28). 4J The confusion matrix for a 2-class problem is given below. Calculate the ,4cc#raey, Sensitivity, Spectf icily and Error rate .. (5) r*J ¢j 2. Predicted Class Class - 1 Class = 0 ActualClass Class - 1 350 150 Class - 0 loo 400 What is the difference between noise and outliers ? Are noise objects always outliers? Are outliers always noise objects ? Justify with examples. What is k-fold cross-validation? How does it differ from the ho]dout method? List two advantages of k-fold cross-validation over the holdout method. Consider the following dataset foi. a 2-class classiflcation problen„ (i) Calculate the Gini index when splitting on attributes A and a. (ji) Which attribute would the decision tree inductioii algorithm choose to split on? (iii)Draw the resulting decision tree showing the number of instances of each class at the leaf nodes. (iv)How many instances are misclassified by the resulting decision tree? 3. Consider the following transaction dataset. Assume the minimum support is 40% and (10) the minimum confidence is 60%. TID Items (i) Show clearly the steps to find frequent itemsets of all sizes using the Apriori algorithm. (i) Generate all the strong association rules from the frequent 3-itemsets. 4. Given the following data points: 3, 6, 8, 14, 16, 22, 28, 32. Assume k = 2, and the initial cluster centroids p1 = 3 and u2 = 14. Show clearly the clusters obtained and the new cluster centres after each iteration of the K-means algorithm for two iterations. 5. The DM Cafe sells cookies with optional toppings: chocolate-chips, walnuts, and (10) oats. Suppose, you have tried five cookies (CI to C5) and kept a record of which ones you liked: Chocolate- Walnuts Oats Liked chips C1 True True False True C2 False True True False True False True True C4 True True True False C5 False False True True