Glmnet varimp fix 2

models/files/glmnet.R

@@ -122,7 +122,7 @@ modelInfo <- list(label = "glmnet",
                       if(length(lambda) > 1) stop("Only one value of lambda is allowed right now")
                       if(!is.null(x$lambdaOpt)) {
                         lambda <- x$lambdaOpt
-                      } else stop("must supply a vaue of lambda")
+                      } else stop("must supply a value of lambda")
                     }
                     allVar <- if(is.list(x$beta)) rownames(x$beta[[1]]) else rownames(x$beta)
                     out <- unlist(predict(x, s = lambda, type = "nonzero"))
@@ -138,14 +138,14 @@ modelInfo <- list(label = "glmnet",
                       if(length(lambda) > 1) stop("Only one value of lambda is allowed right now")
                       if(!is.null(object$lambdaOpt)) {
                         lambda <- object$lambdaOpt
-                      } else stop("must supply a vaue of lambda")
+                      } else stop("must supply a value of lambda")
                     }
                     beta <- predict(object, s = lambda, type = "coef")
                     if(is.list(beta)) {
                       out <- do.call("cbind", lapply(beta, function(x) x[,1]))
                       out <- as.data.frame(out)
                     } else out <- data.frame(Overall = beta[,1])
-                    out <- out[rownames(out) != "(Intercept)",,drop = FALSE]
+                    out <- abs(out[rownames(out) != "(Intercept)",,drop = FALSE])
                     out
                   },
                   levels = function(x) if(any(names(x) == "obsLevels")) x$obsLevels else NULL,

pkg/caret/tests/testthat/test_varImp.R

@@ -0,0 +1,25 @@
+library(caret)
+
+context('Testing varImp')
+
+test_that('glmnet varImp returns non-negative values', {
+  skip_on_cran()
+  skip_if_not_installed('glmnet')
+  set.seed(1)
+  dat <- SLC14_1(200)
+
+  reg <- train(y ~ ., data = dat,
+                      method = "glmnet",
+                      tuneGrid = data.frame(lambda = .1, alpha = .5),
+                      trControl = trainControl(method = "none"))
+
+  # this checks that some coefficients are negative
+  coefs <- predict(reg$finalModel, s=0.1, type="coef")
+  expect_less_than(0, sum(0 > coefs))
+  # now check that all elements of varImp are nonnegative,
+  # in spite of negative coefficients
+  vis <- varImp(reg, s=0.1, scale=F)$importance
+  expect_equal(0, sum(0 > vis))
+})
+
+