Abstract: In order to better guarantee the running effect of the trial run software, the weighted Markov chain is introduced to construct the software defect prediction model. By using the nonlinear computation of weighted Markov chain, the software defect prediction platform based on weighted Markov chain is established, and the relevant parameters affecting the software operation accuracy are calculated within the space scope of the platform, so as to reduce the data dimension in the process of software operation, and improve the accuracy and convergence speed of software defect prediction. The experimental results show that, in the actual application process, the prediction model designed can accurately predict the defects of the test run software, has high practical value, and fully meets the research requirements.

Key words: weighted Markov chain, test run software, defect prediction, prediction platform, data dimension