Human beings always have an urge to predict the future. If one is able to ‘predict’ the future, it is possible to take appropriate action; avoid obstacles and progress in positive direction. The predictions are typically based on historical data. In Software development lifecycle, predictability model or Process Performance models (a term used in CMMi) based on historical data are used to predict various aspects of software.

This paper describes the implementation of predictability model to estimate the number of test cases based on the number of different data element types used on the screen. The business objective is to improve the test coverage so as to improve the product quality and reduce post delivery defects. There were instances where post delivery defects could have been avoided if the relevant test case was present in the test plan. The case study starts with a hypothesis that the number of test cases can be predicted based on various data elements on the screen. Data is collected on the number of data elements and the number of test cases written along with post delivery defects. A regression model is derived based on the data, where dependent variable is number of test cases while independent variables are
different types of data elements on the screen. The regression model is revised based on the improved data. Statistical tests are applied to ensure that the model is a good fit for the data.

The model is now used with the ‘Design’ phase to predict the number of test cases. The predicted number of test cases is given as a target for team member to write the number of test cases. The use of model has resulted in improving the test coverage and reduction of post delivery defects.

Anjali is currently working in Business Excellence team at Atos Origin India. She is responsible for process definition, process improvement and process implementation. She has Master degree in Statistics along with CSQA, PMI, Six Sigma certifications and has an experience in project delivery, quality processes and design of e-learning systems. She is especially interested in carrying out process improvement experiments and implementation on wider scale. Prior to Atos Origin, Anjali has worked at Tata Institute of Fundamental research, Tata Consultancy Services, Silverline, Aptech and Syntel.