Letters in Spatial and Resource Sciences

, Volume 5, Issue 2, pp 55–63

Improving the multi-dimensional comparison of simulation results: a spatial visualization approach

  • Daniel Arribas-Bel
  • Julia Koschinsky
  • Pedro Vasconcelos Amaral
Original Paper

DOI: 10.1007/s12076-011-0064-x

Cite this article as:
Arribas-Bel, D., Koschinsky, J. & Amaral, P.V. Lett Spat Resour Sci (2012) 5: 55. doi:10.1007/s12076-011-0064-x

Abstract

Results from simulation experiments are important in applied spatial econometrics to, for instance, assess the performance of spatial estimators and tests for finite samples. However, the traditional tabular and graphical formats for displaying simulation results in the literature have several disadvantages. These include loss of results, lack of intuitive synthesis, and difficulty in comparing results across multiple dimensions. We propose to address these challenges through a spatial visualization approach. This approach visualizes model precision and bias as well as the size and power of tests in map format. The advantage of this spatial approach is that these maps can display all results succinctly, enable an intuitive interpretation, and compare results efficiently across multiple dimensions of a simulation experiment. Due to the respective strengths of tables, graphs and maps, we propose this spatial approach as a supplement to traditional tabular and graphical display formats.

Keywords

Spatial visualizationMonte Carlo simulation experimentsSpatial econometrics

JEL Classification

Y1C5

Copyright information

© Springer-Verlag 2011

Authors and Affiliations

  • Daniel Arribas-Bel
    • 1
  • Julia Koschinsky
    • 1
  • Pedro Vasconcelos Amaral
    • 1
    • 2
  1. 1.GeoDa Center for Geospatial Analysis and ComputationSchool of Geographical Sciences and Urban PlanningTempeUSA
  2. 2., University of CambridgeCambridgeUK