University of Illinois
College of Business Administration
Department of Information & Decision Sciences
IDS 594 (Special Topics in IDS)   Statistical Finite Mixture Models -- Classification and Cluster Analysis

Fall Semester, 2002
Prof. Stanley L. Sclove 
Homework B   Due Tuesday, 12-November-2002 


Classification
1.     The K = 3 class-conditional densities are exponential with means m =   10, 20 and 30.  (The mean is   1/l, where l is the natural parameter.)
(a)  If there are equal prior probabilities and x = 20, find the probability that x came from the population with mean equal to 20.
(b)  Find the three classification regions,  Rk = {x:  Say Pk}, k = 1,2,3..

Model-Selection for Histograms
2.  For the days ill dataset, use SIC to compare three histograms, one with bins 0, 1, 2, . . . , 18, a second with bins (0),(1,2),(3,4), . . . , (17, 18),  and a third with bins   (0),(1,2,3),(4,5,6),(7,8,9),(10,11,12),(13,14,15),(16,17,18).

K-Means Cluster Analysis
3.  Do a K-Means clustering of the crude oil prices , using a number of clusters and starting points suggested by the histograms from Homework A.  (Use Minitab or other software.)



Copyright (C) 2002     Stanley Louis Sclove
created  2002:  May 29                  latest revision  2002:  Oct 31