Arkadiy Maron – Associate Professor, Department of Business Analytics, Faculty of Business Informatics, National Research University Higher School of Economics.  
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: amaron@hse.ru

Maxim Maron – Director of Technology for Business Ltd., Student of “Business Informatics” MSc Program, Faculty of Business Informatics, National Research University Higher School of Economics.  
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: amaron@hse.ru

To properly execute a project, it is critical to inspect design works for consistency in a timely manner. Otherwise, while it may seem that project schedule is being kept, at some point errors will be found which take much time to correct. To reduce the risk of project schedule disruption, it is advisable to conduct intermediate tests of completed jobs for any errors in a timely manner. At the same time, running to the other extreme should also be avoided: total control would take too much time and money. Therefore, selection of a set of reasonable test points for a project becomes a priority.

The paper suggests a method for setting test points for projects which is based on maximization of information about job number containing an error. The method is based on information approach to testing objects represented by functional models. A modification of information approach to search of possible test point locations is suggested which considers job time length and error localization depth. An optimality criterion is suggested against which comparison will be conducted between sets of test points within projects.

To support  implementation of the suggested approach in practice, a program has been developed for selecting test points for projects which are managed with use of Microsoft Project IT system. A method for conducting experiments to evaluate effectiveness of the suggested approach to test point selection in comparison with a previously developed method which is based on project work time lengths alone is also suggested. To conduct the experiment, special program has been developed which generates source data for two typical structures of project graph: prime sequence and  branched graph.