tests/testthat/test-validateMap.R

In RStoolbox: Remote Sensing Data Analysis

context("validateMap")
lsat_t <- lsat
lsat_t <- lsat_t[[1:4]]

## Set-up test data
set.seed(1)
poly     <- readRDS(system.file("external/trainingPolygons_lsat.rds", package="RStoolbox"))
poly$classNum <- as.numeric(poly$class)

sc <- superClass(lsat_t, trainData = poly, nSamples = 50, responseCol = "class", model = "mlc", trainPartition = 0.7, predict = TRUE)

test_that("classification, without class mapping",{
            val <- validateMap(sc$map, valData = poly, nSample =50, responseCol = "classNum", classMapping = NULL)
            expect_is(val, "mapValidation")
            expect_equal(lapply(val, "class"), list(performance="confusionMatrix",validationSet = "data.frame"))
            expect_equal(colnames(val$validationSet), c("reference", "prediction", "cell"))
        })

test_that("classification, with class mapping",{
            skip_on_cran()
            val <- validateMap(sc$map, valData = poly, nSample = 50, responseCol = "class", classMapping = sc$classMapping)
            expect_is(val, "mapValidation")
            expect_output(print(val), "performance")
            expect_equal(lapply(val, "class"), list(performance="confusionMatrix",validationSet = "data.frame"))
            expect_equal(colnames(val$validationSet), c("reference", "prediction", "cell"))
        })

test_that("regression",{
            skip_on_cran()
            val <- validateMap(sc$map, valData = poly, nSample = 50, mode = "regression", responseCol = "classNum")
            expect_is(val, "mapValidation")
            expect_equal(lapply(val, "class"), list(performance="numeric",validationSet = "data.frame"))
            expect_equal(colnames(val$validationSet), c("reference", "prediction", "cell"))
            expect_equal(names(val$performance)[1:2], c("RMSE", "Rsquared"))
        })