Urban Data Mining Using Emergent SOM

The term of Urban Data-Mining is defined to describe a methodological approach that discovers logical or mathematical and partly complex descriptions of urban patterns and regularities inside the data. The concept of data mining in connection with knowled

  • PDF / 812,270 Bytes
  • 8 Pages / 439.37 x 666.142 pts Page_size
  • 13 Downloads / 174 Views

DOWNLOAD

REPORT


2

Institute of industrial Building Production, University of Karlsruhe (TH), Englerstraße 7, D-76128 Karlsruhe, Germany [email protected] Data Bionics Research Group Philipps-University Marburg, D-35032 Marburg, Germany [email protected]

Abstract. The term of Urban Data-Mining is defined to describe a methodological approach that discovers logical or mathematical and partly complex descriptions of urban patterns and regularities inside the data. The concept of data mining in connection with knowledge discovery techniques plays an important role for the empirical examination of high dimensional data in the field of urban research. The procedures on the basis of knowledge discovery systems are currently not exactly scrutinised for a meaningful integration into the regional and urban planning and development process. In this study ESOM is used to examine communities in Germany. The data deals with the question of dynamic processes (e.g. shrinking and growing of cities). In the future it might be possible to establish an instrument that defines objective criteria for the benchmark process about urban phenomena. The use of GIS supplements the process of knowledge conversion and communication.

1 Introduction Comparisons of cities and typological grouping processes are methodical instruments to develop statistical scales and criteria about urban phenomena. Harris started in 1943, who ranked US cities according to industrial specialization data; many of the other studies that followed added occupational data to the classification models. Later on, in the 1970s, classification studies were geared to measuring social outcomes and shifted more towards the goals of public policy. Forst (1974) presents an investigation of german cities by using social and economic variables. In Great Britain, Craig (1985) employed a cluster analysis technique to classify 459 local authority districts, based on the 1981 Census of Population. Hill et al. (1998) classified US cities by using the city’s population characteristics. Most of the mentioned classification studies use economic, social, and demographic variables as a basis for their classifications which are usually calculated by hierarchical algorithms (e.g. WARD, K-Means). Geospatial objects are analysed by Demsar (2006). These former approaches of city classification are summarized in Behnisch (2007). The purpose of this article is to find groups (clusters) of communities with the same dynamic characteristics in Germany (e.g. shrinking and growing of cities).

312

Martin Behnisch and Alfred Ultsch

The Application of Emergent Self Organizing Maps (ESOM) and the corresponding U*C-Algorithm is proposed for the task of City Classification. The term of Urban Data Mining (Behnisch, 2007) is defined to describe a methodological approach that discovers logical or mathematical and partly complex descriptions of urban patterns and regularities inside the data. The result can suggests a general typology and can lead to the development of prediction models using subgrou