[R-sig-Geo] Prediction of multi-target models on stack objects

[Hugo Machado Rodrigues]

Dear,

I am trying to adjust prediction models for laboratory-measured electrical
conductivity attributes (EC lab) for depths 0-10, 10-30, and 30-50 cm as a
function of apparent electrical conductivity and apparent magnetic
susceptibility attributes (aEC and aMS) measured by the EM38 sensor using
the two coil distances (aEC 1 and 0.5 m; aMS 1 and 0.5 m) in a saline area.

As laboratory attributes have spatial and depth dependency, I would like to
perform prediction models that consider this condition, and for that, I
decided on multi-output (target) modeling methods.

I am using the Elastic Net Regression approach, and I am using the
following piece of code below:

*# Selecting target variables*
*X <- dados_train %>%  select(CE_0_10,CE_10_30,CE_30_50) %>%  scale(center
= F, scale = F) %>% as.matrix()*













*# Selecting predicting variablesY <-dados_train
%>%  select(CE_1_m,CE_0_5_m,SM_1_m,SM_0_5_m) %>% scale(center = F, scale =
F) %> as.matrix()  # Model Building : Elastic Net Regressioncontrol <-
trainControl(method = "repeatedcv", number = 5, repeats = 5, search =
"random", verboseIter = TRUE)  # Training Elastic Net Regression
modelelastic_model_glmnet <- train(CE_0_10+CE_10_30+CE_30_50 ~ .,
                 data = cbind(X, Y),                           method =
"glmnet",                           preProcess = c("center", "scale"),
                     tuneLength = 25,                           trControl =
control)*

My question is:
*How can I predict using the three output models in the stack containing
the EM38 data layers?*

For this, I am using the code below:
# covariates is a stack object containing the 4 rasters of aEC and aMS data
for the two distances (1 and 0.5 m)
*map_glmnet<-raster::predict(covariates,elastic_model_glmnet,index=c(1:3))*

But that doesn't return me a layer for each target attribute (EC lab for
each depth).

Does anyone have any suggestions?

Best Regards

Hugo

	[[alternative HTML version deleted]]