|
|
The project aims at building an R package to implement and develop a published methodology analysing data from order k (k>=2) co-occurrences of events for spatial analysis of multivariate marked processes (point and areal).
kOO === R package: kOO project - start of the project 24/05/2012 - The R package to be built aims at implementing what I did in a few scripts for the preparation of the papers I published in CaGEO and TGIS (see below). Basically this is related to the computation of the distribution of k co-occurrences of spatial events (generalising the contiguity distributions - 2 co-occurrences at distance 0) to derive spatial clustering statistics (mainly using the Shannon entropy, then called the k-spatial entropy) and methods linked to this: SOOk, SelSOOk (see caGEO paper) and scankOO (see TGIS). Another method (CAkOO) performing a k-Correspondence Analysis (i.e. on a multiway table with k entries) on the contingencies of co-occurrenceshas been already "published in my JSS paper about another R package: PTAk. - At the moment the basic scripts I have are based on points and use the R package spatstat, but I would like also to extend to any geometry type. Interested to collaborate on this? === Some details in: === Leibovici, D.G. Bastin, L. Anand, S. Hobona, G. and Jackson, M (2011) "Spatially Clustered Associations in Health related geospatial data". Transactions in GIS, 15(3): 347-364 (June-July 2011) Leibovici D.G. Bastin L,. and Jackson M. (2011) " Higher-Order Co-occurrences for Exploratory Point Pattern Analysis and Decision Tree Clustering on Spatial Data." Computers & Geosciences: 37(3): 382-389 Leibovici, D.G. (2010) "Spatio-temporal Multiway Decomposition using Principal Tensor Analysis on k-modes: the R package PTAk." Journal of Statistical Software, 34(10), 1-34
The project summary page you can find here.