Grupowanie nieruchomości wspólnot mieszkaniowych z wykorzystaniem sieci Kohonena
The clustering of housing community properties with the use of sofm (self-organizing feature maps) method

Summary/Abstract: The following paper presents the results of a housing community clustering research that has been carried out in Olsztyn city. The clustering procedure was realized with the use of the SOFM (self-organizing feature maps) method. The method required particular property features application. These involved: commune share in a housing community, location, land management, condition of the buildings. Having learned and verified the tool a two dimension topological map was made that organized the researched object into eight homogeneous groups. On the basis of the maps and classifying data detailed analysis and cluster synthesis of the given results has been made. The results show that in most of the housing community properties that the Olsztyn commune had a share in, moste of the communes’ shares significantly decreased and the process of scattered privatization was concentrated among properties that had good location and a big need of renovation. The clustered objects differed in terms of their location (general and detailed) and at most medium technical condition. The learned tool is capable of object assessment that has not been presented during the process of learning. The adopted method occurred to be much more effective than the k-average method because if has given much more internally homogenous clusters of the objects. Moreover the given clusters were more transparent and useful in terms of the assumed aim of clustering.

• Details
• Contents