PointPlots using a log scale.

Masked mode

As negative numbers cannot be plotted with logarithms, as a default we just replace them with zeros (maked mode).

Rendering took 24 ms - Size: 35.03 Kb

   data =
      -200..200
      |> Enum.map(fn v -> {v / 10.0, v / 10.0} end)
  
    ds = Dataset.new(data, ["x", "y"])
  
    options = [
      mapping: %{x_col: "x", y_cols: ["y"]},
      data_labels: true,
      orientation: :vertical,
      custom_y_scale:
        ContinuousLogScale.new(domain: {-20, 20}, negative_numbers: :mask),
      colour_palette: ["ff9838", "fdae53", "fbc26f", "fad48e", "fbe5af", "fff5d1"]
    ]
  
    Plot.new(ds, PointPlot, 500, 400, options)
      |> Plot.titles("Masked log scale", "")
      |> Plot.axis_labels("x", "y (log)")
      

Symmetric mode

As negative numbers cannot be plotted with logarithms, we can "make do" and use the symmetric mode.

Rendering took 1 ms - Size: 37.582 Kb

    data =
      -200..200
      |> Enum.map(fn v -> {v / 10.0, v / 10.0} end)
  
    ds = Dataset.new(data, ["x", "y"])
  
    options = [
      mapping: %{x_col: "x", y_cols: ["y"]},
      data_labels: true,
      orientation: :vertical,
      custom_y_scale:
        ContinuousLogScale.new(domain: {-20, 20}, negative_numbers: :sym),
      colour_palette: ["ff9838", "fdae53", "fbc26f", "fad48e", "fbe5af", "fff5d1"]
    ]
  
    Plot.new(ds, PointPlot, 500, 400, options)
      |> Plot.titles("A symmetric log scale", "")
      |> Plot.axis_labels("x", "y (log)")

Linear mode

As numbers below zero are negative as logarithms, and may get really big fast, you may want to "linearize" them.

This works in masked mode (as shown) but also in symmetric mode.

Rendering took 1 ms - Size: 35.118 Kb

  data =
      -200..200
      |> Enum.map(fn v -> {v / 10.0, v / 10.0} end)
  
    ds = Dataset.new(data, ["x", "y"])
  
    options = [
      mapping: %{x_col: "x", y_cols: ["y"]},
      data_labels: true,
      orientation: :vertical,
      custom_y_scale:
        ContinuousLogScale.new(domain: {-20, 20}, negative_numbers: :mask, linear_range: 1),
      colour_palette: ["ff9838", "fdae53", "fbc26f", "fad48e", "fbe5af", "fff5d1"]
    ]
  
    Plot.new(ds, PointPlot, 500, 400, options)
      |> Plot.titles("A masked log scale", "With linear range")
      |> Plot.axis_labels("x", "y")

Automatic range

You can have the logscale "infer" the domain from data, so you don't have to think twice about it.

Rendering took 1 ms - Size: 35.03 Kb

   data =
      -200..200
      |> Enum.map(fn v -> {v / 10.0, v / 10.0} end)
  
    ds = Dataset.new(data, ["x", "y"])
  
    options = [
      mapping: %{x_col: "x", y_cols: ["y"]},
      data_labels: true,
      orientation: :vertical,
      custom_y_scale:
        ContinuousLogScale.new(dataset: ds, axis: "y", negative_numbers: :mask),
      colour_palette: ["ff9838", "fdae53", "fbc26f", "fad48e", "fbe5af", "fff5d1"]
    ]
  
    Plot.new(ds, PointPlot, 500, 400, options)
      |> Plot.titles("Masked log scale", "")
      |> Plot.axis_labels("x", "y (log)")