Urban migration is an important information used in urban planning for traffic, economic and other purposes. Migrations are recorded with origin-destination matrices which give the spatial and temporal components of urban motion. Traditionally, origin-destination matrices are obtained by counting people and vehicles in frequency domains, and in recent times methods used for their development are using data on the use of public mobile (telecommunication) networks for estimating migration. That approach does not report purpose of people’s movements and the reasons for traveling. In this research, a new formal (generalized) method of categorizing spatial objects and predicting probabilities of visits to the categories of these objects is introduced, with intention of enabling the more accurate understanding of the nature of urban migration. The introductory chapter formulates research objectives and a brief overview of the work structure. First chapter details the problem domain and motivation for research. The second chapter describes the data used and their acquisition and processing. In the third chapter the categorization of data was elaborated. The fourth chapter describes the creation of regression model and application on experimental data. The concluding chapter gives the conclusion of the research. Data from the OpenStreetMap spatial database on the use of objects in the Shenzhen city area in China and the GPS data of the taxiway in the same geographic area are used. Spatial objects are categorized by means of the spatial attributes describing objects within OpenStreetMap that describe the purpose of the objects. The categorization is then extended to the taxi data.