Geoscience Reference
In-Depth Information
land use statistics. The latter are currently derived from property registries and
are frequently out-of-date, and updated data is only available upon payment of a
fee (Meinel 2013 ).
3.3
Systematic Approach to Knowledge Discovery in Spatial
Planning Data
As already mentioned, exploratory techniques such as clustering are often employed
in the analysis of spatial planning data. Here we propose to follow a systematic
step-by-step approach, starting from the raw data and ending hopefully with the
discovery of some new data structures, which can then be subjected to rigorous
statistical testing. It should be noted that although the steps are presented here in
succession, in practice it is often the case that insights gained at some step of the
knowledge discovery process will be used to revise the procedures of previous steps.
Therefore, in practice, the process is generally circular or spiral in form, as shown
in Fig. 3.1 (Ultsch 2013 ; Behnisch 2009 ).
The various steps of the current approach to knowledge discovery for spatial
planning data are as follows:
1. Descriptions: Modeling the distribution of each variable separately
-
Initial data inspection
-
Exploring the distributions of the individual variables
2. Structures: Finding structures in high-dimensional space
-
Looking for correlations of variables
V
Knowledge
valid,
comprehensible,
non-trivial,
potentially
innovative and
useful in
practice
Data Inspection
and
Transformation
Knowledge
Generation
Data
DB
o
ss
Expert
System
Domain
Expert
Search WWH ::




Custom Search