Basic Data Mining Questions

Techniques • What is “filter” and “wrapper” method in feature selection? • What’s the advantage and disadvantage of them? • What are the common methods to impute missing data? • What’s the problem associated with learning using imbalance datasets? What are the remedies? Clustering • Explain the bottom-up and top-down approaches in hierarchical clustering. • How to you construct dendrogram from a given data set (HW)? • What is single-link, complete-link, average link, centroid link? • How does k-means algorithm work? • What is SSE in k-means? • How to select the optimal k? • What is a k-mediods method? What’s the advantage of it over k-means? • How to measure clustering quality (internal – scatter criteria, external – precision, recall, …etc ) Apriori Algorithm • What is association rule mining? What are its applications? • What is a strong association rule? • How to calculate support, confidence of a given association rule? • What is the apriori properity? • How to generate C_i’s and L_i’s? • How to evaluate a strong association is interesting or not: lift? What’s the implication of lift >1, = 1, < 1? GSP • What is a sequential pattern? • What is the apriori property in sequence mining? • How to generate the C_i’s and L_i’s? Time series • What are the four components of a time series? • What is an auto-regression model? How to construct lagged predictors? • Why do we need DTW? • What are the 3 properties a warping path must follow? • How to find DTW give a cost matrix?