tests/testthat/test_biplot.R

In mooreryan/biplotr: Give me data, and I'll give you a snazzy biplot!

context("Biplot")
library(ggplot2)
library(grid)
library(biplotr)

test_that("biplot returns a ggplot thing", {
  dat <- matrix(c(1, 2, 3,
                  4, 5, 6,
                  7, 8, 9),
                nrow = 3,
                ncol = 3,
                byrow = TRUE)
  rownames(dat) <- c("apple", "orange", "pear")
  colnames(dat) <- c("price", "amt", "coolness")

  chart <- biplotr::pca_biplot(dat)

  expect_is(chart$biplot, "gg")
})

describe("data subsetting", {
  describe("the user can specify which data columns to keep", {
    it("keeps all columns by default", {

      # Explicitly state the data columns.
      expected <- biplotr::pca_biplot(
        data = iris[, 1:4],
        data_cols = 1:4
      )$pca

      # data_cols defaults to all columsn
      actual <- biplotr::pca_biplot(
        data = iris[, 1:4]
      )$pca

      expect_equal(actual, expected)
    })

    it("handles simple ranges", {
      expected <- biplotr::pca_biplot(
        data = iris[, 1:3]
      )$pca

      actual <- biplotr::pca_biplot(
        data = iris,
        data_cols = 1:3,
      )$pca

      expect_equal(actual, expected)
    })


    it("handles skips in sequential ranges", {
      expected <- biplotr::pca_biplot(
        data = iris[, c(1:2, 4)]
      )$pca

      actual <- biplotr::pca_biplot(
        data = iris,
        data_cols = c(1:2, 4)
      )$pca

      expect_equal(actual, expected)
    })

    it("keeps the original data column order regardless of data_cols ordering", {
      expected <- biplotr::pca_biplot(
        data = iris[, 1:4]
      )$pca

      actual <- biplotr::pca_biplot(
        data = iris,
        data_cols = c(2, 1, 4, 3) # specifying columns in a weird order
      )$pca

      expect_equal(actual, expected)
    })
  })
})

describe("user provides data which isn't all numeric", {
  it("raises an error and gives a decent message", {
    expect_error(
      pca_biplot(iris),
      regexp = "All columns used for PCA are not numeric"
    )
  })
})

describe("coloring data points by group", {
  describe("things that raise errors", {
    it("raises an error if the point_color column name doesn't exist", {
      expect_error(
        pca_biplot(iris,
                   data_cols = 1:4,
                   point_color = "apple_pie"),
        regexp = "point_color column was not present in the data"
      )
    })

    describe("when the data has more covarites than the point_color", {
      it("raises an error unless the data_cols is set properly", {
        # This has Species and apple as covariates.
        iris2 <- cbind(iris, apple = c(rep("a", nrow(iris) / 2), rep("b", nrow(iris) / 2)))

        expect_error(
          # Species was set as point color, but by default it will try and use every column other than point_color as PCA data.
          pca_biplot(iris2, point_color = "Species"),
          regexp = "All columns used for PCA are not numeric"
        )
      })

      it("does NOT raise an error if the extra covarite column is numeric", {
        # This case would be a user error and will give the wrong answer but will not cause an error.
        # In this case, apple is meant to be a covariate and not included in the PCA, but since it is numeric and the default value for data_cols is 1:ncol(data) it IS included in the PCA.
        iris2 <- cbind(iris, apple = 1:nrow(iris))

        expect_error(
          pca_biplot(iris2, point_color = "Species"),
          NA
        )
      })
    })
  })
})

describe("the plot elements that are returned", {
  describe("data_for_ggplot", {
    it("has the PCs and the covarites", {
      p <- pca_biplot(iris, point_color = "Species")

      expected_colnames <- c("PC1", "PC2", "PC3", "Species")
      actual_colnames <- colnames(p$plot_elem$data_for_ggplot)

      expect_equal(actual_colnames, expected_colnames)
    })

    it("returns all the covariates, not just the point_color column", {
      iris2 <- cbind(iris, apple = c(rep("a", nrow(iris) / 2), rep("b", nrow(iris) / 2)))

      p <- pca_biplot(iris2, data_cols = 1:4, point_color = "Species")

      expected_colnames <- c("PC1", "PC2", "PC3", "Species", "apple")
      actual_colnames <- colnames(p$plot_elem$data_for_ggplot)

      expect_equal(actual_colnames, expected_colnames)
    })
  })
})

# Uncomment this section to check on the look of the charts.
describe("coloring data points by group", {
  it("makes a biplot colored by a group", {
    p <- pca_biplot(iris,
                    arrows = TRUE,
                    point_color = "Species",
                    arrow_labels = TRUE,
                    arrow_labels_nudge_y = 0.3,
                    limits_nudge_x = 1,
                    limits_nudge_y = -1)

    pdf("haha.pdf")
    print(p$biplot)
    dev.off()
  })
})