# Cross validation, n-fold and leave-one-out, for 'glmnet' in 'glmnet' package This function is a cross validation function for 'glmnet' method in 'glmnet' package. ```r glmnetcv( trainx, y, family = "gaussian", alpha = 0.5, relax = FALSE, type.measure = "mse", validation = "CV", cv.fold = 10, predacc = "VEcv", ... ) ``` ## Arguments - `trainx`: a matrix contains predictive variables of point samples. The location information, longitude (long), latitude (lat), need to be included in the 'trainx' for spatial predictive modelling. - `y`: a vector of the response variable in the formula, that is, the left part of the formula. - `family`: a description of the error distribution and link function to be used in the model. See '?glmnet' for details. - `alpha,`: an elasticnet mixing parameter, with $0 <= alpha <= 1$. See '?glmnet' for details. - `relax,`: if TRUE then for each active set in the path of solutions, the model is refit without any regularization. See '?glmnet' for more information. - `type.measure,`: loss to use for cross-validation. The default is type.measure="mse". See '?cv.glmnet' for more information. - `validation`: validation methods, include 'LOO': leave-one-out, and 'CV': cross-validation. - `cv.fold`: integer; number of folds in the cross-validation. if > 1, then apply n-fold cross validation; the default is 10, i.e., 10-fold cross validation that is recommended. - `predacc`: can be either "VEcv" for vecv or "ALL" for all measures in function pred.acc. - `...`: other arguments passed on to 'fields'. ## Returns A list with the following components: me, rme, mae, rmae, mse, rmse, rrmse, vecv and e1; or vecv only ## Note This function is largely based on 'glmcv' in this 'spm2' package. ## Examples ```r library(spm) data(petrel) x <- as.matrix(petrel[, c(1, 2, 6:9)]) y <- log(petrel[, 5] + 1) set.seed(1234) glmnetcv1 <- glmnetcv(x, y, validation = "CV", predacc = "ALL") glmnetcv1 data(sponge) x <- as.matrix(cbind(sponge$easting, sponge$easting^2)) set.seed(1234) glmnetcv1 <- glmnetcv(x, sponge[, 3], family = poisson, validation = "CV", predacc = "ALL") glmnetcv1 # For glmnet with gaussian x <- as.matrix(petrel[, c(1, 2, 6:9)]) y <- log(petrel[, 5] + 1) set.seed(1234) n <- 20 # number of iterations,60 to 100 is recommended. VEcv <- NULL for (i in 1:n) { glmnetcv1 <- glmnetcv(x, y, validation = "CV", predacc = "VEcv") VEcv [i] <- glmnetcv1 } plot(VEcv ~ c(1:n), xlab = "Iteration for glmnet", ylab = "VEcv (%)") points(cumsum(VEcv) / c(1:n) ~ c(1:n), col = 2) abline(h = mean(VEcv), col = 'blue', lwd = 2) # For glmnet with binomial x <- as.matrix(cbind(petrel[, c(2, 6)], petrel$long^3, petrel$lat^2, petrel$lat^3)) set.seed(1234) n <- 20 # number of iterations,60 to 100 is recommended. VEcv <- NULL for (i in 1:n) { glmnetcv1 <- glmnetcv(x, petrel[, 5] / 100, family = binomial(link=logit), validation = "CV", predacc = "VEcv") VEcv [i] <- glmnetcv1 } plot(VEcv ~ c(1:n), xlab = "Iteration for glmnet", ylab = "VEcv (%)") points(cumsum(VEcv) / c(1:n) ~ c(1:n), col = 2) abline(h = mean(VEcv), col = 'blue', lwd = 2) ``` ## References A. Liaw and M. Wiener (2002). Classification and Regression by randomForest. R News 2(3), 18-22. ## Author(s) Jin Li

glmnetcv function