Is there a way to implement either a Random-Forest or Lasso model (for variable selection) in R with a pooled spatio-temporal dataset?
Data: My data is a pooled spatial dataset for multiple years. Specifically, it is a polygon with information of properties and when they were sold (I only have one sale per house).
Goal: My goal is to run either a Random Forest or Lasso regression so I can see which variables are important to explain my model.
Issues: I have found some resources but they either:
a) Work with data frames that are not spatio-temporal datasets (example 1, example 2, example 3)
b) Work with spatio-temporal data sets that are raster. And my dataset is a shapefile. Not a raster (such as the CAST package, GeoMLA, example 4, example 5, example 6, example 7, example 8)
I also tried with the spatialRF package, however it is not clear for me how I should define both the distance matrix and the distance threshold matrix.
Any guide would be really appreciated it!
