IBM SPSS Modeler Cookbook

Creating an Outlier report to give to SMEs

It is quite common that the data miner has to rely on others to either provide data or interpret data, or both. Even when the data miner is working with data from their own organization there will be input variables that they don't have direct access to, or that are outside their day-to-day experience.

Are zero values normal? What about negative values? Null values? Are 1500 balance inquiries in a month even possible? How could a wallet cost $19,500? The concept of outliers is something that all analysts are familiar with. Even novice users of Modeler could easily find a dozen ways of identifying some. This recipe is about identifying outliers systematically and quickly so that you can produce a report ...