Question: Consistency is a popular feature subset evaluation measure. The measure is motived by the idea that a good feature subset should have high consistency or
Consistency is a popular feature subset evaluation measure. The measure is motived by the idea that a good feature subset should have high consistency or low inconsistency among matching instances. Two instances are considered matching instances if their values of all features except the class match. For matching instances, an inconsistency occurs if their class values are different. For n matching instances, the number of inconsistencies ie inconsistency count is determined by n maxm m where mmn and m and m are the number instances for class and class respectively. In other words, it is the count of the instances in the minority class. For example, if we only consider the data in the two columns of F and C ignoring features F F and F there are matching instances with F being The inconsistency count for this set of matching instances is max Similarly, there are matching instances with F being and the same inconsistency count The total inconsistency count for F is Consider the following data with four binary features F F F and F and binary class label C Perform feature selection using Sequential Backward Selection SBS search coupled with total inconsistency count as the subset evaluation measure smaller inconsistency is better What will be the selected subset of two features illustrate your solution by intermediate steps F F F F C points For the same data set above, if Sequential forward Selection SFS is used with the same subset evaluation measure, what will be the selected subset of two features illustrate your solution points Discuss which method, SBS or SFS is better for the above data and why?
Step by Step Solution
There are 3 Steps involved in it
1 Expert Approved Answer
Step: 1 Unlock
Question Has Been Solved by an Expert!
Get step-by-step solutions from verified subject matter experts
Step: 2 Unlock
Step: 3 Unlock
