GeoXp-package {GeoXp}R Documentation

Interactive exploratory spatial data analysis

Description

GeoXp is a tool for researchers in spatial statistics, spatial econometrics, geography, ecology etc allowing to link dynamically statistical plots with elementary maps. This coupling consists in the fact that the selection of a zone on the map results in the automatic highlighting of the corresponding points on the statistical graph or reversely, the selection of a portion of the graph results in the automatic highlighting of the corresponding points on the map. GeoXp includes tools from different areas of spatial statistics including geostatistics as well as spatial econometrics and point processes. Besides elementary plots like boxplots, histograms or simple scatterplos, GeoXp also couples with maps Moran scatterplots, variogram cloud, Lorentz Curves,... In order to make the most of the multidimensionality of the data, GeoXp includes some dimension reduction techniques such as PCA.

Details

Package: GeoXp
Type: Package
Version: 1.3
Date: 2008-10-08
License: GPL Vesion 2 or later

The coordinates of sites are represented by two vectors $long$ (for $x-axis$) and $lat$ (for $y-axis$) of size $n$ of numeric values. Attribute value is designed by argument $var$ of size $n$ and can be a vector of numeric values (function histomap, densitymap,...) or a factor (function barmap,...). Some functions use both a vector of numeric values and a factor (function histobarmap, polyboxplotmap,...). Few functions as pcamap and clustermap use a matrix of numeric variables.
Contrary to spdep package (Roger Bivand), a site cannot be represented by lattice/area style. However, user can represent a spatial polygonal contour as background map with option $carte$. A spatial polygonal contour is a matrix of numeric values with 2 columns ($x$ and $y$ coordinates of the vertices of the polygons) where polygons are seperated from each other by 3 rows of NaN. We include in GeoXp 3 functions (polylist2list, spdf2list and map2list) which convert some spatial objects (`Polylist', `SpatialPolygonDataFrame' and `Map' object) into matrix as decribed above to draw a background map.
Among options which are common to each function, users have the possibility to give a $criteria$, vector of boolean of size $n$ with TRUE on specific sites. These sites are then represented by a croice on the map by clicking on `non intercative selection' button.
Moreover, users have the possibility to make bubbles and add some graphs (histogram, barplot or scattermap) giving both a matrix of variables $listvar$ of size $n times p$ and a list of character of size $p$ for variable names. Users can choose a proportional symbol mapping : in function plot, we give value $var^{0.5}$ to argument $cex$. User can choose if a legend has to appear on the map. He could choose then three values represented by bubbles of corresponding sizes.
Finally, users can choose to represent sites on map with different colors (choosed by default and not modifiable by users) in the case of a graph using a factor (argument $color$ with 1 for different colors, 0 otherwise). Users could choose if a legend with corresponding colors will appear on the map. Users can also modify the representation of sites on map (choosed by default and not modifiable by users) with argument $symbol$ (with 1 for different symbols, 0 otherwise).
GeoXp don't use classes yet excepted bar.nb and hist.nb functions which call a nb object (see package spdep). These two functions make an analysis of a neighborhood matrix.

Author(s)

Christine Thomas-Agnan, Yves Aragon, Anne Ruiz-Gazen, Thibault Laurent, Laurianne Robidou Maintainer: Thibault Laurent <thibault.laurent@univ-tlse1.fr>

References

Aragon Yves, Perrin Olivier, Ruiz-Gazen Anne, Thomas-Agnan Christine (2008), ``Statistique et Econométrie pour données géoréférencées : modèles et études de cas''


[Package GeoXp version 1.3 Index]