Investigation of non-spatial and spatial methods of analysis
The goal of this study was to identify the similarities and differences in methods of analysis for spatial and non-spatial data. The two methods used to analyze spatial and non-spatial data are kernel density estimation and clustering. The first part of the thesis discusses the use of the two different methods on each of the data sets. The latter portion discusses the use of spatial renditions to the data and the comparison of the methods. By comparing the different methods on the different types of data, the differences in spatial and non-spatial analysis can be determined.
The first chapter of the thesis discusses the differences in spatial and non-spatial data. This includes the introduction of the term spatial autocorrelation and the use of topology. Chapter two discusses the use the kernel density and clustering methods on the non-spatial micro array data and the results of the analysis. Chapter three examines the use of the same methods in chapter two used on pancreatic cancer data. These methods do not include the use of spatial autocorrelation in the data. Chapter four investigates the use of spatially rendered kernel density and clustering methods. These spatially rendered models include the use of spatial autocorrelation in analysis. Chapter four also discusses the use of different exploratory spatial data analysis methods to analyze the data. Chapter five is the discussion of the results from the study.
The results from chapters two and three allowed for comparison of the kernel density and clustering methods for spatial and non-spatial data. Similarities and differences in the methods can be determined from chapters two and three. Chapter four allows for the comparison of the spatially rendered and original methods on the pancreatic cancer data. Comparing these results showed the difference in including spatial autocorrelation.
The results of the study indicated the difference in using the kernel density and clustering on the spatial and non-spatial data. The two methods were inadequate for analyzing spatial data. Thus the need for spatially rendered methods is shown. The use of spatial kernel density estimation and spatial clustering is also supported by the use of additional exploratory spatial data analysis methods.