Use missing instead of nothing in DataFrame() overload

docs/src/examples/dataframes.md

@@ -29,7 +29,7 @@ using Statistics
 using DataFrames
 
 # Load environmental data
-temperature, precipitation = worldclim([1,12])
+temperature, precipitation = SimpleSDMPredictor(WorldClim, BioClim, [1,12])
 
 # Get GBIF occurrences
 kingfisher = GBIF.taxon("Megaceryle alcyon", strict=true)
@@ -59,11 +59,12 @@ rename!(env_df, :x1 => :temperature, :x2 => :precipitation)
 first(env_df, 5)
 ```
 
-Note that the resulting `DataFrame` will include the values set to `nothing` in
-the layers. We might want to remove those rows using `filter!`:
+Note that the resulting `DataFrame` will include `missing` values for the 
+elements set to `nothing` in the layers. We might want to remove those rows 
+using `filter!` or `dropmissing!`:
 
 ```@example dataframes
-filter!(x -> !isnothing(x.temperature) && !isnothing(x.precipitation), env_df);
+dropmissing!(env_df, [:temperature, :precipitation]);
 last(env_df, 5)
 ```
 

src/SimpleSDMLayers.jl

@@ -72,11 +72,11 @@ export clip
 function __init__()
     @require GBIF="ee291a33-5a6c-5552-a3c8-0f29a1181037" begin
         @info "GBIF integration loaded"
-        include(joinpath(dirname(pathof(SimpleSDMLayers)), "integrations", "GBIF.jl"))
+        include("integrations/GBIF.jl")
     end
     @require DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0" begin
         @info "DataFrames integration loaded"
-        include(joinpath(dirname(pathof(SimpleSDMLayers)), "integrations", "DataFrames.jl"))
+        include("integrations/DataFrames.jl")
     end
 
 end

src/integrations/DataFrames.jl

@@ -54,19 +54,24 @@ end
     DataFrames.DataFrame(layer::T) where {T <: SimpleSDMLayer}
 
 Returns a DataFrame from a `SimpleSDMLayer` element, with columns for latitudes,
-longitudes and grid values. 
+longitudes and grid values. `nothing` values in the layer grid are replaced by
+`missing` values in the DataFrame.
 """
 function DataFrames.DataFrame(layer::T; kw...) where {T <: SimpleSDMLayer}
     lats = repeat(latitudes(layer), outer = size(layer, 2))
     lons = repeat(longitudes(layer), inner = size(layer, 1))
-    return DataFrames.DataFrame(latitude = lats, longitude = lons, values = vec(layer.grid); kw...)
+    values = replace(vec(layer.grid), nothing => missing)
+    df = DataFrames.DataFrame(longitude = lons, latitude = lats, values = values; kw...)
+    return df
 end
 
 """
     DataFrames.DataFrame(layers::Array{SimpleSDMLayer})
 
 Returns a single DataFrame from an `Array` of compatible`SimpleSDMLayer`
-elements, with every layer as a column, as well as columns for latitudes and longitudes.
+elements, with every layer as a column, as well as columns for latitudes and 
+longitudes. `nothing` values in the layer grids are replaced by
+`missing` values in the DataFrame.
 """
 function DataFrames.DataFrame(layers::Array{T}; kw...) where {T <: SimpleSDMLayer}
     l1 = layers[1]
@@ -75,10 +80,11 @@ function DataFrames.DataFrame(layers::Array{T}; kw...) where {T <: SimpleSDMLaye
     lats = repeat(latitudes(l1), outer = size(l1, 2))
     lons = repeat(longitudes(l1), inner = size(l1, 1))
     values = mapreduce(x -> vec(x.grid), hcat, layers)
+    values = replace(values, nothing => missing)
 
     df = DataFrames.DataFrame(values, :auto; kw...)
-    DataFrames.insertcols!(df, 1, :latitude => lats)
     DataFrames.insertcols!(df, 1, :longitude => lons)
+    DataFrames.insertcols!(df, 2, :latitude => lats)
     return df
 end
 

test/dataframes.jl

@@ -5,22 +5,33 @@ using Test
 
 temperature = SimpleSDMPredictor(WorldClim, BioClim, 1)
 
-df = DataFrame(latitude = [0.0, 1.0], longitude = [30.0, 31.0], values = [42.0, 15.0])
+df = DataFrame(longitude = [30.0, 31.0], latitude = [0.0, 1.0], values = [42.0, 15.0])
 
 @test eltype(temperature[df]) <: Number
 
 temperature_clip = clip(temperature, df)
 
 @test typeof(temperature_clip) == typeof(temperature)
 
-@test typeof(DataFrame(temperature_clip)) == DataFrame
-@test eltype(DataFrame(temperature_clip).values) == eltype(temperature_clip.grid)
-@test typeof(DataFrame([temperature_clip, temperature_clip])) == DataFrame
-@test eltype(DataFrame([temperature_clip, temperature_clip]).x1) == eltype(temperature_clip.grid)
+df1 = DataFrame(temperature_clip)
+df2 = DataFrame([temperature_clip, temperature_clip])
+
+@test typeof(df1) == DataFrame
+@test eltype(df1.values) == Union{Missing, eltype(temperature_clip)}
+@test typeof(df2) == DataFrame
+@test eltype(df2.x1) == Union{Missing, eltype(temperature_clip)}
 
 @test typeof(SimpleSDMPredictor(df, :values, temperature_clip)) <: SimpleSDMLayer
 @test typeof(SimpleSDMPredictor(df, :values, temperature_clip)) <: SimpleSDMPredictor
 
+l1 = SimpleSDMPredictor(df1, :values, temperature_clip)
+l2 = SimpleSDMPredictor(df2, :x2, temperature_clip)
+for l in (l1, l2)
+    @test isequal(l.grid, temperature_clip.grid)
+    @test isequal(longitudes(l), longitudes(temperature_clip))
+    @test isequal(latitudes(l), latitudes(temperature_clip))
+end
+
 mbool = mask(temperature_clip, df, Bool)
 @test eltype(mbool) == Bool