Bump deps 0.9 (#65)

CategoricalArrays 0.9 Distributions 0.24 FillArrays v0.10 TableOperations v1
Nosferican · Dec 7, 2020 · f742839 · f742839 · Nosferican · Dec 7, 2020
1 parent b58bcaf
commit f742839
Show file tree

Hide file tree

Showing 8 changed files with 55 additions and 58 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,2 +1,3 @@
 docs/build/
 Manifest.toml
+scratch/
diff --git a/Project.toml b/Project.toml
@@ -3,7 +3,7 @@ uuid = "4d6a76a9-bfbc-5492-8924-cf6ed7875f06"
 license = "ISC"
 description = "Econometrics in Julia"
 maintainer = "José Bayoán Santiago Calderón"
-version = "0.2.7"
+version = "0.2.8"
 
 [deps]
 CategoricalArrays = "324d7699-5711-5eae-9e2f-1d82baa6b597"
@@ -22,18 +22,18 @@ TableOperations = "ab02a1b2-a7df-11e8-156e-fb1833f50b87"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 
 [compat]
-CategoricalArrays = "0.8"
-Distributions = "0.23"
-FillArrays = "0.9"
+CategoricalArrays = "0.8, 0.9"
+Distributions = "0.23, 0.24"
+FillArrays = "0.9, 0.10"
 ForwardDiff = "0.10"
-MLJBase = "0.15"
+MLJBase = "0.15, 0.16"
 MLJModelInterface = "0.3"
 Optim = "1"
 Parameters = "0.12"
 StatsBase = "0.33"
 StatsFuns = "0.9"
 StatsModels = "0.6"
-TableOperations = "0.2"
+TableOperations = "1"
 Tables = "1"
 julia = "1.4"
 

diff --git a/docs/jmd/estimators.jmd b/docs/jmd/estimators.jmd
@@ -163,11 +163,10 @@ model = fit(RandomEffectsEstimator,
 ## Nominal Response Models
 
 ```julia; line_width = 90;
-data = joinpath(dirname(pathof(Econometrics)), "..", "data", "insure.csv") |>
-   CSV.File |> DataFrame |>
+data = CSV.read(joinpath(pkgdir(Econometrics), "data", "insure.csv"), DataFrame) |>
    (data -> select(data, [:insure, :age, :male, :nonwhite, :site])) |>
    dropmissing |>
-   (data -> categorical!(data, [:insure, :site]))
+   (data -> transform!(data, [:insure, :site] .=> categorical, renamecols = false))
 ```
 
 The model automatically detects that the response is nominal and uses the correct model.
@@ -185,7 +184,7 @@ model = fit(EconometricModel,
 data = RDatasets.dataset("Ecdat", "Kakadu") |>
        (data -> select(data, [:RecParks, :Sex, :Age, :Schooling]))
 data.RecParks = convert(Vector{Int}, data.RecParks)
-data.RecParks = levels!(categorical(data.RecParks, ordered = true), collect(1:5))
+data.RecParks = levels!(categorical(data.RecParks, ordered = true, compress = true), collect(1:5))
 data
 ```
 

diff --git a/src/Econometrics.jl b/src/Econometrics.jl
@@ -166,5 +166,6 @@ export @formula,
     HC1,
     HC2,
     HC3,
-    HC4
+    HC4,
+    levels!
 end
diff --git a/test/linear_models.jl b/test/linear_models.jl
diff --git a/test/mlj.jl b/test/mlj.jl
@@ -1,8 +1,7 @@
 @testset "MLJ model interface" begin
-    data = data = CSV.File(joinpath(Econometrics.DATAPATH, "insure.csv"), select = [:insure, :age, :male, :nonwhite, :site]) |>
-        DataFrame |>
+    data = CSV.read(joinpath(pkgdir(Econometrics), "data", "insure.csv"), DataFrame, select = [:insure, :age, :male, :nonwhite, :site]) |>
         dropmissing |>
-        (data -> categorical!(data, [:insure, :site]))
+        (data -> transform!(data, [:insure, :site] .=> categorical, renamecols = false))
     @testset "Econometrics.jl + MLJModelInterface.jl" begin
         @testset "specify contrasts" begin
             contrasts = Dict(:insure => DummyCoding(base = "Uninsure"))

diff --git a/test/mlogit.jl b/test/mlogit.jl
@@ -1,9 +1,8 @@
 # Nominal Models
 @testset "Nominal Models" begin
-    data = CSV.File(joinpath(Econometrics.DATAPATH, "insure.csv"), select = [:insure, :age, :male, :nonwhite, :site]) |>
-        DataFrame |>
+    data = CSV.read(joinpath(pkgdir(Econometrics), "data", "insure.csv"), DataFrame, select = [:insure, :age, :male, :nonwhite, :site]) |>
         dropmissing |>
-        (data -> categorical!(data, [:insure, :site]))
+        (data -> transform!(data, [:insure, :site] .=> categorical, renamecols = false))
     model = fit(
         EconometricModel,
         @formula(insure ~ age + male + nonwhite + site),
@@ -26,7 +25,7 @@
         0.7577178,
         1.324599,
         -0.3801756,
-    ] rtol = 1e-3
+        ] rtol = 1e-3
     @test V ≈
           [
         0.35084518 -0.00590456 -0.02519209 -0.01837512 -0.08995794 -0.08264268 0.29928937 -0.00510107 -0.02257270 -0.01585492 -0.07392512 -0.06808314
@@ -41,7 +40,8 @@
         -0.01585492 -0.00024122 0.01369872 0.15068111 0.01594691 -0.02353016 -0.02105510 -0.00028680 0.01625685 0.17604391 0.02099458 -0.03008391
         -0.07392512 0.00033847 0.00945927 0.01664999 0.19895500 0.05182727 -0.08991231 0.00039990 0.01101870 0.02099458 0.22070774 0.06273376
         -0.06808314 0.00035236 0.00802593 -0.02345786 0.05189742 0.11060487 -0.08165509 0.00041790 0.00922333 -0.03008391 0.06273376 0.13899387
-    ] |> Hermitian rtol = 1e-3
+        ] |>
+        Hermitian rtol = 1e-3
     @test σ ≈ [
         0.5923219,
         0.0114418,
@@ -55,5 +55,5 @@
         0.4195759,
         0.4697954,
         0.3728188,
-    ] rtol = 1e-4
+        ] rtol = 1e-4
 end
diff --git a/test/ologit.jl b/test/ologit.jl
@@ -4,7 +4,7 @@
         dataset("Ecdat", "Kakadu") |>
         (data -> select(data, [:RecParks, :Sex, :Age, :Schooling]))
     data.RecParks = convert(Vector{Int}, data.RecParks)
-    data.RecParks = levels!(categorical(data.RecParks, ordered = true), collect(1:5))
+    data.RecParks = levels!(categorical(data.RecParks, ordered = true, compress = true), collect(1:5))
     model = fit(EconometricModel, @formula(RecParks ~ Age + Sex + Schooling), data)
     @test sprint(show, model) ==
           "Probability Model for Ordinal Response\nCategories: 1 < 2 < 3 < 4 < 5\nNumber of observations: 1827\nNull Loglikelihood: -2677.60\nLoglikelihood: -2657.40\nR-squared: 0.0075\nLR Test: 40.42 ∼ χ²(3) ⟹ Pr > χ² = 0.0000\nFormula: RecParks ~ Age + Sex + Schooling\n──────────────────────────────────────────────────────────────────────────────────────────\n                          PE          SE        t-value  Pr > |t|        2.50%      97.50%\n──────────────────────────────────────────────────────────────────────────────────────────\nAge                  0.00943647  0.00254031    3.71469     0.0002   0.00445424   0.0144187\nSex: male           -0.0151659   0.0846365    -0.179188    0.8578  -0.181161     0.150829\nSchooling           -0.103902    0.0248742    -4.17711     <1e-4   -0.152687    -0.0551174\n(Intercept): 1 | 2  -2.92405     0.191927    -15.2352      <1e-48  -3.30047     -2.54763\n(Intercept): 2 | 3  -1.54922     0.171444     -9.03632     <1e-18  -1.88547     -1.21297\n(Intercept): 3 | 4  -0.298938    0.166904     -1.79108     0.0734  -0.626281     0.0284051\n(Intercept): 4 | 5   0.669835    0.167622      3.9961      <1e-4    0.341083     0.998587\n──────────────────────────────────────────────────────────────────────────────────────────"
@@ -17,7 +17,7 @@
         -1.549266200,
         -0.298963900,
         0.6698249,
-    ] rtol = 1e-4
+        ] rtol = 1e-4
     @test V ≈ [
         0.00000646 -0.0000007000 0.000013250 0.00031473 0.0003161358 0.000321276 0.0003270734
         -0.00000070 0.0071633800 -0.000184930 0.00294759 0.0029647442 0.002995366 0.0029904485
@@ -26,7 +26,7 @@
         0.00031614 0.0029647400 0.002854820 0.02892393 0.0293971651 0.027160433 0.0266006675
         0.000321276 0.0029953660 0.002791361 0.02698623 0.0271604331 0.027860143 0.0270590982
         0.0003270734 0.002990448 0.002731893 0.02649876 0.0266006675 0.027059098 0.0281002038
-    ] rtol = 1e-3
+        ] rtol = 1e-3
     @test σ ≈ [
         0.002540789,
         0.084636770,
@@ -35,37 +35,34 @@
         0.171456015,
         0.166913580,
         0.167631154,
-    ] rtol = 1e-4
-    data = data[data.RecParks.≠5, :]
+        ] rtol = 1e-4
+    data = data[data.RecParks .≠ 5, :]
     model = fit(EconometricModel, @formula(RecParks ~ Age + Sex + Schooling), data)
     @test sprint(show, model) ==
           "Probability Model for Ordinal Response\nCategories: 1 < 2 < 3 < 4\nNumber of observations: 1200\nNull Loglikelihood: -1502.61\nLoglikelihood: -1500.73\nR-squared: 0.0013\nLR Test: 3.76 ∼ χ²(3) ⟹ Pr > χ² = 0.2886\nFormula: RecParks ~ Age + Sex + Schooling\n───────────────────────────────────────────────────────────────────────────────────────────\n                          PE          SE        t-value  Pr > |t|        2.50%       97.50%\n───────────────────────────────────────────────────────────────────────────────────────────\nAge                  0.00063565  0.00329173    0.193105    0.8469  -0.00582255   0.00709385\nSex: male            0.0896101   0.106068      0.84484     0.3984  -0.118489     0.297709\nSchooling           -0.0509934   0.0300878    -1.69482     0.0904  -0.110024     0.00803731\n(Intercept): 1 | 2  -2.58123     0.228989    -11.2723      <1e-27  -3.0305      -2.13197\n(Intercept): 2 | 3  -1.12053     0.212192     -5.28071     <1e-6   -1.53684     -0.704217\n(Intercept): 3 | 4   0.486702    0.210541      2.31167     0.0210   0.0736316    0.899772\n───────────────────────────────────────────────────────────────────────────────────────────"
     β, V, σ = coef(model), vcov(model), stderror(model)
-    @test β ≈ [0.00063565, 0.08961012, -0.05099339, -2.5812334, -1.1205278, 0.48670172] rtol =
-        1e-4
+    @test β ≈ [0.00063565, 0.08961012, -0.05099339, -2.5812334, -1.1205278, 0.48670172] rtol = 1e-4
     @test V ≈ [
         0.00001084 -5.386e-06 0.00002186 0.00052578 0.0005259 0.00052967
         -5.386e-06 0.01125032 -0.00021416 0.00465658 0.00470423 0.00484966
         0.00002186 -0.00021416 0.00090528 0.00435314 0.00431928 0.00424373
         0.00052578 0.00465658 0.00435314 0.05243606 0.04424607 0.04221716
         0.0005259 0.00470423 0.00431928 0.04424607 0.04502565 0.0424849
         0.00052967 0.00484966 0.00424373 0.04221716 0.0424849 0.04432745
-    ] rtol = 1e-3
+        ] rtol = 1e-3
     @test σ ≈ [0.0032917, 0.1060675, 0.0300878, 0.2289892, 0.2121925, 0.2105408] rtol = 1e-4
-    data =
-        joinpath(dirname(pathof(Econometrics)), "..", "data", "auto.csv") |> CSV.File |> DataFrame |>
-        (data -> select(data, [:rep77, :foreign, :length, :mpg])) |> dropmissing |>
-        categorical!
-    data.rep77 = levels!(
-        categorical(data.rep77; ordered = true),
+    data = CSV.read(joinpath(pkgdir(Econometrics), "data", "auto.csv"), DataFrame) |>
+        (data -> select(data, [:rep77, :foreign, :length, :mpg])) |>
+        dropmissing
+    data.foreign = categorical(data.foreign, compress = true)
+    data.rep77 = levels!(categorical(data.rep77; ordered = true, compress = true),
         ["Poor", "Fair", "Average", "Good", "Excellent"],
-    )
+        )
     model = fit(EconometricModel, @formula(rep77 ~ foreign + length + mpg), data)
     @test sprint(show, model) ==
           "Probability Model for Ordinal Response\nCategories: Poor < Fair < Average < Good < Excellent\nNumber of observations: 66\nNull Loglikelihood: -89.90\nLoglikelihood: -78.25\nR-squared: 0.1295\nLR Test: 23.29 ∼ χ²(3) ⟹ Pr > χ² = 0.0000\nFormula: rep77 ~ foreign + length + mpg\n─────────────────────────────────────────────────────────────────────────────────────────────\n                                    PE         SE     t-value  Pr > |t|       2.50%    97.50%\n─────────────────────────────────────────────────────────────────────────────────────────────\nforeign: Foreign                2.89681    0.790641   3.66387    0.0005   1.31684     4.47678\nlength                          0.0828275  0.02272    3.64558    0.0005   0.0374253   0.12823\nmpg                             0.230768   0.0704548  3.2754     0.0017   0.0899749   0.37156\n(Intercept): Poor | Fair       17.9275     5.55119    3.22948    0.0020   6.83431    29.0206\n(Intercept): Fair | Average    19.8651     5.59648    3.54956    0.0007   8.68139    31.0487\n(Intercept): Average | Good    22.1033     5.70894    3.8717     0.0003  10.6949     33.5117\n(Intercept): Good | Excellent  24.6921     5.89075    4.19168    <1e-4   12.9204     36.4639\n─────────────────────────────────────────────────────────────────────────────────────────────"
     β, V, σ = coef(model), vcov(model), stderror(model)
-    @test β ≈ [2.89679875, 0.08282676, 0.23076532, 17.92728, 19.86486, 22.10311, 24.69193] atol =
-        1e-3
+    @test β ≈ [2.89679875, 0.08282676, 0.23076532, 17.92728, 19.86486, 22.10311, 24.69193] atol = 1e-3
     @test V ≈ [
         0.62578335 0.0111044038 0.012321411 2.4421745 2.4836507 2.5750058 2.6957587
         0.0111044 0.0005186744 0.001222385 0.1243671 0.1258979 0.1284727 0.1319772
@@ -74,7 +71,7 @@
         2.48365067 0.1258979463 0.336805731 31.0279194 31.4396158 31.9986912 32.9170873
         2.57500582 0.1284726666 0.343025653 31.6005528 31.9986912 32.7044311 33.6189243
         2.69575871 0.1319772302 0.354971081 32.5099187 32.9170873 33.6189243 34.8533724
-    ] rtol = 1e-2
+        ] rtol = 1e-2
     @test σ ≈ [
         0.79106469,
         0.02277442,
@@ -83,5 +80,5 @@
         5.60710405,
         5.71877881,
         5.90367448,
-    ] rtol = 1e-2
+        ] rtol = 1e-2
 end