Customer Value Prediction - Real Clients' Data Analysis

The aim of this bachelor's thesis is to find such relationships which would contribute to submitters and owners of data or a sponsoring company, all by using the real clients' data analysis. The analysis will be performed based on the CRISP-DM methodology. As for the analytical techniques, CF-Miner and ETree-Miner, which are the part of the LISp-Miner system, are used. The thesis is divided into two parts - theoretical and practical. In the theoretical part, we will learn about the process of the knowledge discovery in databases and what general processes exist. Further, there is the description of the particular phases of the KDD process with the respect to chosen methodology. At the end of this part, we will learn more about the academic system LISp-Miner which serves for the knowledge discovery in databases; this system is based on the GUHA method. Finally, there is the description of the analytical procedures CF-Miner and ETree-Miner which are being used in the analysis. The practical part is structured in compliance with the CRISP-DM methodology. At first, the scope of analysis and the data used in the analysis are introduced. Then there is the explanation of the phase of the data preparation, this phase precedes the simulation itself. In the scope of the simulation phase, the particular analytical tasks are being solved. The whole analysis process was very time demanding, mainly due to the data preparation. However, the data have appeared as less appropriate and thus there are suggestions for improvement listed in the last chapter.