Introduction

You should have a data set that contains at least three \(\mathrm{X}\) variables. The \(\mathrm{X}\) variables can be continuous and/or categorical variables. In the case of a classification/discrimination problem, you should identify one additional variable that identifies two or more populations(classes) into which you wish to be able to classify any observation vector on your X variables. You should use clustering either to identify classes or populations into which you wish to sort your observations or to verify the reasonableness of the classes you already have. You should then do a discrimination analysis using an appropriate method for linear discrimination to develop one or more discrimination function using your variables. You should then use either additional data or partitioned test set to estimate the actual error rate(AER) of misclassification.

We have discussed cluster analysis briefly and classification/discriminant analysis in more detail. You should use this information to form clusters for your data, either to give populations for classification or to verify that the existing populations make sense. You should then do linear discrimination analysis(using \(80 / 20\) partition split or step-wise discrimination if you choose), checking at least roughly for normality and transforming variables as needed, checking for equal covariance matrices for the different groups, finding the discrimination rules and using at least the partitioned test set to estimate the AER of misclassification.

It should contain a 2-3 page summary of the problem your model attempts to solve, the data set used, the final model and your main findings. The focus here is to write a coherent narrative that states the high points of what you found in your project. If necessary, include appendices that give the most important printouts or reproduces parts of those printouts of any computer programs or spreadsheets used to do the analysis

Price: $5.21

Solution: The downloadable solution consists of 3 pages, 221 words.
Deliverable: Word Document