Industrial Case Study on Time Series Analysis of Metrics Changes Based on GQM Models

It is difficult for companies to evaluate whether their activities have achieved the intended goals or to improve their goals in quantifiable ways. Herein we propose a method to evaluate quantitatively the results of a company's activities related to goals based on the GQM (Goal Question Metrics) method. Our proposed method uses successive statistical analysis and periodic statistical analysis. The former evaluates two groups of projects based on project release date, and the latter evaluates projects separated by six-month intervals. An empirical case study for a cloud service, which supports real estate businesses with e-Seikatsu, is used to evaluate our method. The target data contain project quality evaluation results for three years. We evaluated 73 project quality evaluation results using the proposed method. Successive statistical analysis reveals four of the five metrics have statistical significance of mean differences between groups, whereas periodic statistical analysis indicates three of the five metrics are statistically significant.