Request: how to generate contingency matrices to evaluate thematic rasters

[hariasa]

Hi everyone,

I was wondering if anyone knows a handy and efficient way to create a contingency matrix, evaluating a thematic raster layer (result from a classification) with the source polygons (in either vector or raster). This will be handy to get the users accuracy and producers accuracy and to specify whether a classification makes sense or not.

I know it is possible to do this in ERDAS when actually classifying, but I am looking for a method that enables me to do it after the results have been already generated.

I was thinking in a way of converting polygons to rasters with arcGIS, and then summarizing specific raster values together in order to fill in the contingency matrix in i.e. Excel, but this seems like a lot of work. Does anyone know a faster way of doing so? Perhaps in R Statistics?

I don't know much about programming yet so any help here would be awesome. Thanks! 

[Lurker]

so, you want to correlate between thematic layers from classification with the source?

I just thinking a Pearson's Chi-squared test, but of course you need to populate it manually,

regards,

[nomlas]

Hi

Since you are aware of the classification accuracy measures available in Erdas (Envi also has some nice post classification accuracy measures like AUC and ROC) and Lurker has mentioned the Chi squared test another option will be to use the map comparison  kit available at  -->Map Comparison Kit Website,

For details and examples, please see: H. Visser and T. de Nijs, 2006. The Map Comparison Kit. Environmental Modeling & Software 21, 346-358.

Thanks