(CoKriging Spatio-Temporal interpolate))

CoSTK





 
A short description 
The distribution package 
Related publications 
Authors 
Contact 


A short description

Managing geophysical data generated by the emerging spatio-temporal data sources (e.g. geosensor networks) presents a growing challenge to GISience. The presence of correlation (i.e. spatial correlation across several geosensor sites and time correlation within each site) poses difficulties with respect to traditional spatial data analysis. This system integrates a novel spatio-temporal analytical scheme that allows us to yield a characterization of correlation in geophysical data along the spatial and temporal dimension. It resorts to a multivariate statistical model, namely CoKriging, in order to derive accurate spatio-temporal interpolates, which predict unknown data by utilizing not only own geosensor values at the same time, but also information from both near past and future data. It uses a window-based computation methodology that leverages the power of temporal correlation in a spatial modeling phase by also fitting the computed interpolate to data which may change over time.
S. Pravilovic, A. Appice, D. Malerba. Leveraging Correlation across Space and Time to Interpolate Geophysical Data via CoKriging. Under Review (March 2017)

The distribution package

The CoSTK algorithmic pipeline (as well as its spatial (Kriging) and spatio-temporal competitors (STKriging) is implemented in a R environment (distributed through the CRAN network).

code/data Description
Data sets This zip bundle contains dataset directories, which collect various geophysical data collections
Kriging This script (written in R) imports a data set: (1)the geosensor network (for each geosensor, the identifier and its 2D spatial coordinate) and (2) the spatio-temporal data matrix (i.e. a data matrix with Z(T,G) corresponding to the datum observed at time T from geosensor G). It computes the Ordinary Kriging interpolation models on the k-fold CV of the selected data set. It outputs the computed accuracy results.
STKriging This script (written in R) imports a data set. It computes the STKriging interpolation models based on the k-fold CV of the data set. It outputs the computed accuracy results.
CoSTK This script (written in R) imports a data set. It computes the CoSTK interpolation models based on the k-fold CV of the data set. It uses the window model and considers the top ranked Principal Component computed over the windowed data raws as the secondary co-variable. It outputs the computed accuracy results.
CoKriging This script (written in R) imports a data set. It computes the CoKriging interpolation models based on the k-fold CV of the data set. It uses the window model and considers one secondary co-variable for each windowed data raw. It outputs the computed accuracy results.
CoSTK90 This script (written in R) imports a data set. It computes the CoSTK interpolation models based on the k-fold CV of the data set. It uses the window model and considers Principal Components explaining the 90% of variability in windowed data raws as secondary co-variables. It outputs the computed accuracy results.
ReadMe
Kriging This script (written in R) imports a data set: (1)the geosensor network (for each geosensor, the identifier and its 2D spatial coordinate) and (2) the spatio-temporal data matrix (i.e. a data matrix with Z(T,G) corresponding to the datum observed at time T from geosensor G). It computes the Ordinary Kriging interpoaltion models on the entire selected data set. It outputs the computation times.
STKriging This script (written in R) imports a data set. It computes the STKriging interpolation model of the entire data set. It outputs the computation times.
CoSTK This script (written in R) imports a data set. It computes the CoSTK interpolation models of the entire data set. It uses the window model and considers the top ranked Principal Component computed over the windowed data raws as the secondary co-variable. It outputs the computation times.
CoKriging This script (written in R) imports a data set. It computes the CoKriging interpolation models of the entire data set. It uses the window model and considers one secondary co-variable for each windowed data raw. It outputs the computation times.
CoSTK90 This script (written in R) imports a data set. It computes the CoSTK interpolation models of the entire data set. It uses the window model and considers Principal Components explaining the 90% of variability in windowed data raws as secondary co-variables. It outputs the computation times.
ReadMe

Warning: The CoSTK software is free for evaluation, research and teaching purposes, but not for commercial purposes. The software is provided "as is" without warranty of any kind, either expressed or implied.
 
 

Top of this page

Authors

  • Sonja PRAVILOVIC
  • Annalisa APPICE
  • Donato MALERBA


    Contact

    Sonja Pravilovic sonja.pravilovic@unimediteran.net, sonja.pravilovic@uniba.it
    Annalisa Appice annalisa.appice@uniba.it
     

    Top of this page