dropmissingsubj

Project.toml

@@ -3,7 +3,7 @@ uuid = "00e1d80e-504f-4d7b-946f-0aea2e7abfe3"
 keywords = ["bioequivalence"]
 desc = "Bioequivalence"
 authors = ["Vladimir Arnautov <[email protected]>"]
-version = "0.2.3"
+version = "0.2.4"
 
 [deps]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"

src/MetidaBioeq.jl

@@ -4,7 +4,7 @@ using  MetidaNCA, Metida, GLM, MixedModels, DataFrames, CategoricalArrays, Distr
 
 import Base: show
 
-export result, bioquivalence, estimate
+export result, bioquivalence, estimate, makeseq
 
 include("types.jl")
 include("bioequivalence.jl")

src/bioequivalence.jl

@@ -33,6 +33,7 @@ end
         seqcheck::Bool = true,
         designcheck::Bool = true,
         dropcheck::Bool = true,
+        dropmissingsubj = false,
         info::Bool = true,
         warns::Bool = true,
         autoseq::Bool = false,
@@ -49,6 +50,7 @@ end
 * `seqcheck` - check sequencs; 
 * `designcheck` - check design correctness;  
 * `dropcheck` - dropuot check;
+* `dropmissingsubj` - drop subjects with missing data;
 * `info` - show information;
 * `warns` - show warnings;
 * `autoseq` - try to make sequence collumn;
@@ -67,6 +69,7 @@ function bioequivalence(data;
     seqcheck::Bool = true,
     designcheck::Bool = true,
     dropcheck::Bool = true,
+    dropmissingsubj::Bool = false,
     info::Bool = true,
     warns::Bool = true,
     autoseq::Bool = false,
@@ -188,7 +191,25 @@ function bioequivalence(data;
         if dropcheck
             if !isnothing(vars) && !nomissing(data, vars)
                 info && @info "Dropuot(s) found in dataframe!"
-                dropout = true
+                if dropmissingsubj
+                    droplists = map(x-> Set(data[findall(ismissing, data[!, x]), subject]), vars)
+                    if all(x-> isequal(x, first(droplists)), droplists)
+                        sbjdel   = x -> x ∉ first(droplists)
+                        data     = filter(subject => sbjdel, data)
+                        subjects = filter(sbjdel, subjects)
+
+                        subjnum  = length(subjects)
+                        obsnum   = size(data, 1)
+                        info && @info "Dropuot(s) removed!"
+                        dropout = false
+                    else
+                        warns && @warn "Different subjects have missing values for different variables. Dropuot(s) NOT removed!"
+                        dropout = true
+                    end
+
+                else
+                    dropout = true
+                end
             elseif !isnothing(vars)
                 info && @info "No dropuot(s) found in dataframe!"
                 dropout = false

test/tests.jl

@@ -6,6 +6,8 @@ io       = IOBuffer();
 
 rdsdict    = Dict()
 rdsdict[1] = rds1 = bedf2x2 = CSV.File(joinpath(path, "csv", "2x2rds1.csv")) |> DataFrame
+
+
 transform!(bedf2x2, :Subj => categorical, renamecols = false)
 transform!(bedf2x2, :Per => categorical, renamecols = false)
 bedf2x2.logVar = log.(bedf2x2.Var)
@@ -16,6 +18,13 @@ transform!(bedf2x2x4, :Period => categorical, renamecols = false)
 
 refvals = CSV.File(joinpath(path, "csv", "ciref.csv")) |> DataFrame
 
+
+rds1missing = allowmissing!(deepcopy(rds1), [:Var, :logVar])
+rds1missing[1, :Var] = missing
+rds1missing[36, :Var] = missing
+rds1missing[1, :logVar] = missing
+rds1missing[36, :logVar] = missing
+
 @testset "  Basic test" begin
 
     @test MetidaBioeq.nomissing(bedf2x2x4, :logVar) == false
@@ -68,8 +77,6 @@ refvals = CSV.File(joinpath(path, "csv", "ciref.csv")) |> DataFrame
     @test beres.method  == "B"
     @test beres.estimator== "mm"
 
-
-
     # Crossover design 2X2
     #
     be2 = MetidaBioeq.bioequivalence(bedf2x2, 
@@ -168,7 +175,48 @@ refvals = CSV.File(joinpath(path, "csv", "ciref.csv")) |> DataFrame
     @test_nowarn  MetidaBioeq.estimate(be2;  estimator = "mm", method = "B")
     @test_nowarn  MetidaBioeq.estimate(be2;  estimator = "met", method = "B")
 
-end
+    # Drop missing
+    # no drop
+    bedm = MetidaBioeq.bioequivalence(rds1missing, 
+    vars = [:Var, :logVar], 
+    subject = :Subj, 
+    formulation = :Trt, 
+    reference = "R",
+    period = :Per,
+    sequence = :Seq, 
+    autoseq = true)
+
+    @test bedm.dropout
+    @test length(bedm.subjects) == 18
+
+    bedm = MetidaBioeq.bioequivalence(rds1missing, 
+    vars = [:Var, :logVar], 
+    subject = :Subj, 
+    formulation = :Trt, 
+    reference = "R",
+    period = :Per,
+    sequence = :Seq, 
+    autoseq = true,
+    dropmissingsubj = true)
+
+    @test bedm.dropout == false
+    @test length(bedm.subjects) == 16
+
+    rds1missing[2, :Var] = missing
+    @test_warn "Different subjects have missing values for different variables. Dropuot(s) NOT removed!" bedm = MetidaBioeq.bioequivalence(rds1missing, 
+    vars = [:Var, :logVar], 
+    subject = :Subj, 
+    formulation = :Trt, 
+    reference = "R",
+    period = :Per,
+    sequence = :Seq, 
+    autoseq = true,
+    dropmissingsubj = true)
+
+    @test bedm.dropout
+    @test length(bedm.subjects) == 18
+
+end # end Basic tests
 
 rdsdict[2] = CSV.File(joinpath(path, "csv", "2x2rds2.csv")) |> DataFrame
 rdsdict[3] = CSV.File(joinpath(path, "csv", "2x2rds3.csv")) |> DataFrame