Projects in IIT



B.Tech Project

Automatic Job Modeling Using Clone Detection Algorithm

  • A user submitted job is characterised by Before Execution properties(BE data) and After Execution properties(AE data).

  • BE data are input size, cyclomatic complexity, number of functions, loops etc. AE data are execution time, processor time, memory space etc. We store BE and AE data of several submitted jobs in MySQL database.

  • When a user submits a job, we extract the BE data of it, using which we use metric based clone detection algorithm to match to which of the previously submitted jobs, it is similar.

  • Then, we fetch the AE data of those similar tasks and apply linear or multiple regression to predict the AE data of the newly submitted job.

  • This technique can be applied in a job scheduling algorithm in a distributed system.