Dobývání znalostí z databází na datech z trhu ojetých a skladových automobilů značky BMW

The bachelor's thesis deals with the application of the CRISP-DM methodology to data from the Czech market of used and stock BMW vehicles, with the aim of identifying factors influencing pricing. The data were obtained using the Scrapy tool from the TipCars portal and subsequently subjected to a process of cleaning, transformation, and analysis. Exploratory analysis provided an overview of the offered models, technical characteristics of the vehicles, and their geographical distribution. In the analytical phase, a Random Forest Regression model was applied, which achieved acceptable prediction accuracy and highlighted the year of registration, power, and engine capacity as key predictors. The DBSCAN clustering method revealed natural groups of vehicles based on technical parameters, with some corresponding to sports versions, others to regular models or electric cars. The results of the work can be used to support decision-making in the secondary car market. The limitations of the research include subjective assessment of technical condition, absence of equipment data and uneven representation of categories.