Variable Selection - Essay Example

Variable selection Variable selection Bruce Ratner prefers forward and backward selection because they are better selection methods as opposed to stepwise. Bruce Ratner views stepwise selection as being far too narrow in yielding confidence and will not produce good results if used when the data has redundant predictors. This is opposed to forward and backward selection which are very efficient in such situations and many others (Bruce, 2011). Many other people do not share the same opinion as Bruce Ratner: many computer data analysts find stepwise selection to be very efficient and good to use.

They prefer when the stepwise selection method is used in the algorithm of data analysis programs such as SAS and SPSS. Based on my personal experience I think the author is correct, forward and backward selection produces better results in varied conditions as opposed to stepwise selection.A good selection technique that I have in mind will involve the use of estimation of empirical samples through the use of repeated sampling of data samples. This algorithm allows for the approximation of the distributed test statistics will be usable in small scale data where the large scale results may not hold.

This algorithm is efficient because it helps solve the mistake of automated variable selection methods. There are other designed algorithms that are employed under various situations. One that is interesting is the Naïve Bayes which is based on the Bayes theorem. I think it is not an efficient algorithm because it generalizes samples which easily occurs to errors in analysis of the data samples.ReferencesBruce Ratner (2011) Statistical and Machine Learning Data Mining: Techniques for Better Predictive Modeling and Analysis of Big Data 2nd Edition. CRC Press