Poisson Smoother

Like the empirical Bayesian smoother the Poisson smoothing method is a tool for dealing with rate instability associated with small sample sizes.  In SpaceStat, the smoother takes a rate and a population dataset and performs a kriging analysis using the Poisson population adjustment.  The result is two datasets: an estimate dataset (the smoothed rate) and an estimate variance dataset.  Note that you can perform the same analysis manually using the kriging method with a Poisson variogram model.

Input Panel

The Input Panel for the Poisson smoother is identical to the input panel for Empirical Bayesian.

 

Rate dataset: Specify the rate dataset to be smoothed.

Population dataset:  Specify an associated population dataset defining the population counts for each object.

Start time: The start time of the analysis

End time: The end time of the analysis

Spatial weight set: Specify a spatial weight set defining each object's neighbors to be used in generating a smoothed rate.  The poisson smoother requies spatial weight sets which include the ego.

Units multiplier: The denominator of the population dataset.  If the population values are "per 100000 individuals" then the units multiplier should be 100000.

Advanced Panel

This panel allows you to use areal deconvolution in your smoothing.

Polygon geographies using areal deconvolution produce a more appropriate variogram model for kriging.  Instead of calculating covariance between neighboring polygons based on their centroid distance, multiple points in each polygon are generated and the average of the covariances of the point pairs is used.

Output Panel

This panel provides the option to customize the name of the variogam model which is produced by this method.

You can examine this model after the smoothing analysis completes to insure the model fits the data reasonably well.  It will appear in the Variogram model view after the method has run.