Knowledge Discovery in Spatial Planning Data: A Concept for Cluster Understanding - Computational Approaches for Urban Environments - page 67

coastal urban district, for example, the historical settlement development, the
shape of the coastline, shape of the bay, size of the harbor area, development of
traffic infrastructure, municipal environmental protection, proximity effects of the
surrounding area, natural and administrative boundaries, as well as the location of
settlement functions. In some cases, coastal waters can affect the development of
built-up areas and traffic infrastructure in coastal urban districts, which in turn will
have an impact on open space fragmentation. Such information could be used for
further spatial investigations of coastal urban district types.
Another interesting thing was, for example, the identification of a cluster that was
characterized by many protected areas and a compact building utilization (e.g., small
value of building area per inhabitant). The cluster objects also belong to the first
discovered group of urban districts with lower values of sealed surfaces (see Fig. 3.3 :
Exploring the distributions of the individual variables). The rule describing Cluster
UC4 is:

UD data belongs to Cluster UC4, if
sqrt . ProtectedAreas / 60:8555 and
log . BuildingArea / < 75:0076 and
log . SealedSurfaces / < 48:3179

Freiburg im Breisgau was found to be representative for this cluster. This may
lead to another hypothesis: There is a type of urban districts in Germany the ones
with an assumed efficient and compact building structure and a prominent influence
of natural protection laws on the settlement development. The following label is
suggested: “Urban Protected Areas”. The details and in particular the causes of
the observed land use properties (e.g., lower values of sealed surfaces, small value
of building area per inhabitant) can presumably not be exhaustively ascertained
on the basis of the presented approach. In addition to the quantitative approach,
a qualitative examination of selected results might be necessary (e.g., integration
of case studies, interviews, local inspections). One should ask to what degree the
settlement structure or the type of buildings can be traced back to conscious planning
decisions of local decision-makers. Moreover, the question as to the extent to which
local construction activity is regulated by regional spatial planning is of interest.
However, this is not an issue of the presented quantitative investigation.
In summary, an investigation of clusters in the ways sketched above may lead
to a fresh interpretation of what features of the data are of interest. In this way,
the application of exploratory techniques can eventually produce knowledge that is
entirely new.

3.4 Conclusions and Future Challenges

In view of the continual growth in geodata and the heterogeneity of datasets,
there is an urgent need for alternatives to traditional spatial analysis within
geographic information systems (GIS) (Miller and Han 2009 , p. 13). Techniques