Analysis of real consumer loans data by LISp-Miner system

The goal of this paper is to find interesting relations in real consumer loans data, which could be used practically. The analysis is done using LISp-Miner system which is a software designed for data discovery in databases based on the GUHA method. The analysis will use specifically 4ft-Miner and SD4ft-Miner procedures described below. The methodology used is CRISP-DM. The contribution of my work lies in finding unknown or confirming known relations in the field of consumer loans, which could be used practically. Further contribution lies in presentation of possibilities of data discovery in databases and the non-commercial LISp-Miner system. The bachelor thesis consists of a theoretical and a practical part. In the three chapters assigned for the theoretical part I explain at first, what is data discovery in databases, when it appears, why is it interesting and its techniques with an emphasis on the association rules used in this paper. Afterwards I briefly describe the LISp-Miner system and demonstrate the procedures used on artificial data. In the last theoretical chapter I introduce methodologies of data discovery in databases with an accent on CRISP-DM, which is the one used here. The practical part corresponds with the individual steps of the CRISP-DM metodology, i.e. business understanding, data understanding, data preparation, modeling, evaluation and deployment. Because of clarity I decided to join last two steps into one.