# Regression Analysis with single value as predictor and grid as dependant variable

I'd like to perform a regression analysis in QGIS. Let's say for example that I have 3 grids as my dependent variable and 3 single values as my predictor. That is, for a single value of the predictor I'll have as many values as dependent variable as the number of cells in a grid. For another single value as predictor, I'll have as many values as dependent variable as the number of cells of another grid and so on. I want to perform a regression analysis between the predictor and the dependent, for example with a power law such as `Y=a*(X^b)` where `X` is a single value and `Y` all the values of a single grid. As a result, I'll have consequently two grids as output, one for `a` values and one for `b` values.

Is it possible to do so, maybe using Python?

## 2 Answers

Can you clarify: do you want a single number to predict the value everywhere on a grid? That is, is the grid a constant value?

Yes i want the single number to predict the value everywhere but the grid is not constant. Let's say for example, i have 3 single values and 3 grids, with 4 cell each one. All the grids have the same extension. Given a spatial location (that is, given a single cell in a certain position), i'll have 3 dependant values, one for each grid, in that cell: each one of these 3 values is related to a single value of the predictors. If i do the regression, i'll obtain "a" and "b" parameters valid for the cell i've considered. I move to another cell, repeat the process and obtain "a" and "b" parameters valid for this second cell, and so on. What i'll have eventually is 4 values for "a", and 4 for "b". I hope that I have been able to explain the problem. Thanks for your answer!

• Might be easier if you edit your original question and show a map of your data and the grids. Tell us which features are inputs and which are outputs. The SAGA plugin has regression analysis for points and predictor grids, but I think you are doing the reverse. – wingnut Apr 7 at 5:04