Proposal of effective preprocessing techniques of financial data

The article focuses on describing process of customization data obtained from ERP software. The aim of this customization was ensuring usability of data collection for obtaining knowledge from data via methods of data mining. First part describes importance of preprocessing phase of working with big data sets. It deals with terms commonly used in the topic - methodology CRISP-DM and its phases, with closer look at individual steps of preprocessing. Second part contains description of the chosen dataset, extraction of raw data and their required format (according to their expected future use). The data are financial indications of a real concern. This part also focuses on methods and programs used for preprocessing the data. Third part corresponds with the main task of this article, what is adjusting obtained data to format which would be appropriate to further processing and analyzing purposes with Data Mining techniques. It describes five phases of preprocessing, modified in a way needed for manipulating with chosen data. The last part compiles acquired knowledge and presents further use of the data after transformation from raw form to demanded format. The paper concludes with BPMN diagram of the process with focusing on preprocessing (data preparation) stage.