For any spatial data, you can calculate Moran's I, which determines the presence of spatial autocorrelation. This can be done in ArcGIS, Geoda, and R software.

You may go through the below link for determining Moran's I using Geoda: http://geodacenter.github.io/workbook/5a_global_auto/lab5a.html#the-moran-scatter-plot

Sowanjaya is correct. On a more broad basis, you should also not only check consumption, but other statistical model factors:

Have you accounted for the following? Relative poverty, level of and type of power generation, and overall sufficiency of data? If you don't have a large population from which to derive consumption of energy data, you will get gaps, and that may require something like Kriging.

Best of success to you!

In ArcGiS spatial autocorrelation function is available. But you need to follow some instructions like you have raster data right need to convert it to the feature class then integrate the data. After integration you can use the collective events tools to show the distribution of feature classes. Then directly you can use getis-ord-Zi and next Moran-1 spatial autocorrelation function. All you have found in the Arctool box. The data should be arranged from the vector to raster format and do data projection as per your Datum.

Mohammad Jaber

The data you need to calculate the presence of spatial autocorrelation is a spatial weights matrix, which you thinks suits you the most and the variable in which you want to check the spatial autocorrelation. The method calculates spatial autocorrelation in the dependent variable. However, recent literature calculates the spatial autocorrelation in the residuals. In this case, you replace the dependent variable with the residuals that you get from a linear regression model.

To detect spatial correlation you need a Shape database. GeoDa is easy to use and allow getting basic information. As Mouni Maddela says it, you need to define a weight matrix; you try many specifications and choose the more appropriate one. The weight matrix is used to calculate the Moran’s indicator and to test its significance (other tests of spatial autocorrelation are available; the Moran’s one is the most used). When SA detected, the use of local indicators, bivariate Moran’s scatter plot is suitable before going to spatial modelization.

In simple words.

For example, Region X (district) is divided into 500 smaller spatial units (sub-districts). Now you should have two basic data to run spatial autocorrelation: i. Shapefile (.shp format) and ii. attribute data (values of the variable of interest for all 500 smaller units). For these kinds of study, you have to go for Local Moran's I. This can be performed in ArcGIS as well as in Geoda. For lesser complexity go for Geoda (open source).

Steps....

1. open the shapefile

2. save your excel datasheet into geoda supported formate (.CSV)

3. go to the table bar on geoda and merge this table with your shapefile.

4. now click the W symbol to generate spatial weights. Save the weight file.

5. go to Space tab and select the Univariate local moran I option.

6. select all three results option for better interpretation.

Similarly, if you want to establish a relation between two variables, opt for Bivariate Local Moran I.

Thank you everyone for your valuable answers!