When Regression Fails: Combining JMP® Techniques to Handle Nonconforming Data
Cheryl E. Johnson, AVP and Operations Research Analyst II, Bank of America
A technology team at Bank of America wished to use historical data to predict the performance of new projects based on the combination of the applications being tested for programming bugs. This team is tasked with designing and running test scripts to identify bugs so that they can be corrected prior to putting any applications with updates or new functionality back into production. Data Analytics & Modeling supports this team and was asked to develop the predictive model. The objective of the analysis was two-fold: (1) initially estimate count and type of bugs and the timeframe in which they would likely to be detected, and (2) predict whether a project already underway could be completed on-time based on issues which had already been identified and the expected length of time to resolve them