The growth of complexity in modern applications arises new problems to the operators of these systems. In order to clearly understand their applications, they often use Application Performance Monitoring tools. These tools, by their design, gather a lot of data from the application. This data holds valuable information, but users only provided with a fraction of it.
Machine learning went through enormous development in the past years, and expected to revolutionize many work processes in the future. I suppose, that machine learning methods could help us discover new relationships in our data, which was gathered by Application Performance Monitoring tools.
In my thesis I attempt to find a method to extract more information from the data gathered by Application Performance Monitoring tools. I find a scalable solution for extracting and storing the data. Then I transform and analyze my dataset, and at last I build a predictive model, which is sufficient for making predictions on future records.