I need a tool in spatial statistic tool which will help me to calculate error statistically between 3 data sets(taking one as standard).

I need a tool in spatial statistic tool which will help me to calculate error statistically between 3 data sets(taking one as standard).

Dear sir

i have three polygon, having little daviate from one another. I want to

calculate that deviation statistically, so suggest me some tool from

spatial statistics which can help me to cal culate the daviation(error).

On Wed, Feb 17, 2016 at 5:21 AM, Chris Donohue, GISP <geonet@esri.com>

Hi Litan

If you have access to the

**Spatial Analyst**toolbox, one approach may be toconvert each polygon to its own raster, then use the *Band Collection

Statistics* tool to calculate the statistics for the set of 3 rasters. Make

sure to tick the optional

**Compute covariance and correlation matrices**option.

The output is text file providing various statistics. The correlation

matrix is useful in that you can now compare (via correlation) how similar

each of the rasters are to one-another (and your "standard" layer).

For a quick whirlwind tour of every tool in the

**Spatial Statistics**toolbox, do watch this tech session video from last year's Esri User

Conference

http://video.esri.com/watch/4702/simple-ways-to-do-more-with-your-data-using-spatial-statistics

Good luck.

Mervyn

- 1 person found this helpful
If you are trying to compare 2 polygons whether in raster or in vector form. There is no statistical test that will return a p-value. Formulating the hypothesis that the polygons are different would need to be ascribed to some phenomenon. Does that phenomenon describe how the polygons would differ? For example:

- Would they differ by shape?
- By size?
- is there a directional difference?
- does the shape of the boundary change with respect to the delineating points?
- do the shapes move?

So in short, if you find that there is a difference, you can do subtraction (the equivalent in raster and vector format) and quantify the difference.

Also, for something to be tested significantly, you need a threshold sample/population size. You say you have 2 or at most 3 polygons. Saying that they are significantly different or not is not possible since any observations... and hence conclusions... are drawn from limited observations.

In short... simply state they are different... state the amount of the difference... and leave the conjecture or extrapolation of conclusion to the reader.

- 1 person found this helpful
Please start using the title field as a title field. Long lists of comma delimited words and verbose chatter in the title are driving away folks who might answer your questions. If you want your questions answered, I recommend you work harder towards framing a complete question, with a description of the software in use, and a complete description of your inputs and desired outputs. Only after the question is complete (and it should have a question mark to indicate the issue, framed as a question) should you go back and design a title which reflects the core issue. A good title will attract the interest of folks who have the knowledge you seek. In this case, maybe "Statistical error determination across three datasets".

Spatial statistics An overview of the Spatial Statistics toolbox—Help | ArcGIS for Desktop and

Geostatistical analyst An overview of the Geostatistical Analyst toolbar and toolbox—Help | ArcGIS for Desktop

are your two built in options. Are you looking for a particular test? If you have one that is a standard, the suggestions would depend upon the form of the data that you have. Is it raster? vector? Is the standard a standard at a point in time? More information would be useful