I have two gridcell-based point layers that I am interested in investigating their spatial correlation in QGIS, or alternatively in R as well. The variables I am interested in are:

  1. Map 1: Point layer of percent change in price
  2. Map 2: Point layer of demand estimate

A few word on the context of the analysis. The maps are generated via a simple supply-demand pricing framework, which is spatially explicit at the gridcell level. For each gridcell, we have a supply curve that characterizes the availability of a resource k based on data on availability and procurement cost. Using scenarios from a GIS-based model, I can generate different demand schedules, of which Map 2 is an example. First, we use a business-as-usual scenario to calibrate a BAU price vector. Then, changing certain constraint parameters in the GIS-based model, we run alternative scenarios which generate a different demand profile, and for which we generate their specific price vectors using the supply curve. Thus, the price impacts represent the percent change deviation from the BAU and the alternative scenario of which Map 1 is an example.

Hence, what I am interested in is to somehow evaluate the degree of correlation between the demand location and the price impact location. My questions are: What is the best way to go about this? Is it possible to do correlation analysis in QGIS? Or is best to run spatial regressions to test the fit between the price impact and demand layers?

I have looked at certain alternatives in QGIS. Specifically, I have managed to perform a Nearest Neighbour Analysis (NNA) to investigate the linkages between the location of the price impacts in Map 1 and the location of the demand in Map 2. However, I wonder if that is the 'best' route to go about.

Any advice and/or pointers would be very welcome.

Map 1 Map 2

closed as unclear what you're asking by Spacedman, aldo_tapia, whyzar, Kersten, BERA Nov 22 '17 at 13:23

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    Can you please clarify exactly what you mean by "spatial correlation"? There are statistics aimed at describing the spatial process, clustering from a random CSR process and actual correlation. This correlation can be performed within a specified distance of the same data or across data at a fixed location. These correlations are normally not "spatial" per se. One exception is the Dutilleul (1993) modified t-test that corrects degrees of freedom based on autocorrelation. What have you tried? Have you looked at the literature? – Jeffrey Evans Nov 21 '17 at 15:59
  • Sorry for the late reply, but I have been trying to do my homework on the field of spatial economics, or more spatial econometrics. I must say I am fairly new to the field and fairly unfamiliar with the literature. Hence the nature of my question about getting pointers about which directions I should be heading. I have found method that are GIS-based such the Nearest Neighbour Analysis, however I am not sure if that is a suitable framework to follow given the nature of my problem. I have clarified the question a bit further by giving a brief description of my problem. – iouraich Nov 29 '17 at 11:29
  • 1
    I think this is a very useful question. Essentially how can I tell if and how much these two datasets are correlated? – HeikkiVesanto Nov 30 '17 at 16:11