3

I am trying to find spatial correlation between two variables , and in my case i have two kind of data and they are : number of populations and number of schools , my goal is how to find a relation between those data using Moran indice or if you have any suggest to do such thing.

My data are like this

here is my shape file and data

  • 1
    What isyour data? Do you have a point feature class for schools as well? So far, what I see is a set of administrative units that has two attributes. Are you trying to get the correlation between the attributes Population and School? If that is the case, we are looking at basic statistical correlation. Spatial correlation measures clustering or dispersion: desktop.arcgis.com/en/arcmap/10.3/tools/… – Fabien Ancelin Mar 30 '17 at 14:37
  • thanks for rely , Yes i have point feature of the schools too, i am looking to get index value to represent the relation between school / population relation – user943998 Mar 30 '17 at 14:46
  • 1
    What would be your hypotheses in testing the bivariate autocorrelation? This is usually applied to change in the same process and not two different processes and is not remotely the same as correlation. The Moran's-I describes the spatial structure. If you want to look at correlation structure then use a statistic such as the Pearson's product-moment correlation coefficient. – Jeffrey Evans Mar 30 '17 at 21:26
4

From what I understand, computing a simple correlation between your two attributes in your administrative areas would give you that: one coefficient that tells you the relationship between the two attributes for your data set.

Moran index is designed to solve clustering problematic: http://pro.arcgis.com/en/pro-app/tool-reference/spatial-statistics/spatial-autocorrelation.htm

Edit:

R exposes capabilities to compute this: http://www.r-tutor.com/elementary-statistics/numerical-measures/correlation-coefficient

I know there there are some ways to used R with ArcGIS but I have never used it, so I you want to go that way, you will have to explore that on your own: https://blogs.esri.com/esri/arcgis/2017/02/03/r-arcgis-news-and-plans-for-2017/

  • ok , how calculating this coefficient , should i divide population bu schools for exemple ? if so how do they have spatial relation ? – user943998 Mar 30 '17 at 15:04
  • No. You use the Pearson coefficient correlation which is based on the covariance between the 2 variables. A coefficient of -1 means perfect inverse correlation whereas + 1 means perfect positive correlation. Excel can compute that for you using the formula CORREL but I don't know of any way to do it in ArcMap. It is not complicated to implement a tool to do that either but the formula, though simple, it bit complicated to explain and wiki does that well: en.wikipedia.org/wiki/Correlation_and_dependence – Fabien Ancelin Mar 30 '17 at 15:11
  • Actually, this links explains you that even better is you want to code it: thoughtco.com/… – Fabien Ancelin Mar 30 '17 at 15:18
  • The Morna's-I is a statistic for describing global spatial structure of a random field and is not for intended as a clustering statistic. Although, the coefficient does describe different types of spatial association, it is not always in terms of clustering. It is more in terms of attraction and diffusion processes, following the two tails of the expected Moran's distribution. – Jeffrey Evans Mar 30 '17 at 21:29
  • Thank you for explain , I will use Pearson coefficient correlation, and i will try to make it sense and relate with a map. – user943998 Apr 1 '17 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.