fix doctests and add missing docstrings (#192)

* first round of doctest(; fix=true): coeftables, hypotheiscoding

* use Documenter 0.25 series; docstrings for response/modelmatrix

* include hypothesis_matrix in contrasts doc page

* DataFrames.DataFrame -> DataFrame

* build docs on 1.5

* bump patch version

* specify rng in internals

* add StableRNGs to project

* stablerng everywhere there's a rand

* don't need rand here

* don't seed RNG in setdocmeta

* run doctest(; fix=true)

* missed some fixes...

* use julia-repl for the ugly hypmat text and add some explanatory txt

* bump patch version

* fix a few more stray small exponents

* Revert "fix a few more stray small exponents"

This reverts commit c458cdf.

Author: kleinschmidt

Project.toml

@@ -1,6 +1,6 @@
 name = "StatsModels"
 uuid = "3eaba693-59b7-5ba5-a881-562e759f1c8d"
-version = "0.6.16"
+version = "0.6.17"
 
 [deps]
 DataAPI = "9a962f9c-6df0-11e9-0e5d-c546b8b5ee8a"

docs/Project.toml

@@ -2,9 +2,10 @@
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 GLM = "38e38edf-8417-5370-95a0-9cbb8c7f171a"
+StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 StatsModels = "3eaba693-59b7-5ba5-a881-562e759f1c8d"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 
 [compat]
-Documenter = "~0.24"
+Documenter = "0.25"

docs/make.jl

@@ -1,5 +1,8 @@
 using Documenter, StatsModels
 
+DocMeta.setdocmeta!(StatsModels, :DocTestSetup, :(using StatsModels, StatsBase); recursive=true)
+
+
 using Pkg
 Pkg.precompile()
 
@@ -12,7 +15,8 @@ makedocs(
         "Contrast coding categorical variables" => "contrasts.md",
         "Temporal variables and Time Series Terms" => "temporal_terms.md",
         "API documentation" => "api.md"
-    ]
+    ],
+    modules = [StatsModels]
 )
 
 deploydocs(

docs/src/contrasts.md

@@ -51,6 +51,7 @@ EffectsCoding
 HelmertCoding
 SeqDiffCoding
 HypothesisCoding
+hypothesis_matrix
 ```
 
 ### Special internal contrasts

docs/src/formula.md

@@ -2,8 +2,6 @@
 CurrentModule = StatsModels
 DocTestSetup = quote
     using StatsModels
-    using Random
-    Random.seed!(1)
 end
 ```
 
@@ -42,6 +40,8 @@ Here is an example of the `@formula` in action:
 ```jldoctest 1
 julia> using StatsModels, DataFrames
 
+julia> using StableRNGs; rng = StableRNG(1);
+
 julia> f = @formula(y ~ 1 + a + b + c + b&c)
 FormulaTerm
 Response:
@@ -53,20 +53,20 @@ Predictors:
   c(unknown)
   b(unknown) & c(unknown)
 
-julia> df = DataFrame(y = rand(9), a = 1:9, b = rand(9), c = repeat(["d","e","f"], 3))
-9×4 DataFrames.DataFrame
-│ Row │ y          │ a     │ b         │ c      │
-│     │ Float64    │ Int64 │ Float64   │ String │
-├─────┼────────────┼───────┼───────────┼────────┤
-│ 1   │ 0.236033   │ 1     │ 0.986666  │ d      │
-│ 2   │ 0.346517   │ 2     │ 0.555751  │ e      │
-│ 3   │ 0.312707   │ 3     │ 0.437108  │ f      │
-│ 4   │ 0.00790928 │ 4     │ 0.424718  │ d      │
-│ 5   │ 0.488613   │ 5     │ 0.773223  │ e      │
-│ 6   │ 0.210968   │ 6     │ 0.28119   │ f      │
-│ 7   │ 0.951916   │ 7     │ 0.209472  │ d      │
-│ 8   │ 0.999905   │ 8     │ 0.251379  │ e      │
-│ 9   │ 0.251662   │ 9     │ 0.0203749 │ f      │
+julia> df = DataFrame(y = rand(rng, 9), a = 1:9, b = rand(rng, 9), c = repeat(["d","e","f"], 3))
+9×4 DataFrame
+ Row │ y          a      b         c
+     │ Float64    Int64  Float64   String
+─────┼────────────────────────────────────
+   1 │ 0.585195       1  0.236782  d
+   2 │ 0.0773379      2  0.943741  e
+   3 │ 0.716628       3  0.445671  f
+   4 │ 0.320357       4  0.763679  d
+   5 │ 0.653093       5  0.145071  e
+   6 │ 0.236639       6  0.021124  f
+   7 │ 0.709684       7  0.152545  d
+   8 │ 0.557787       8  0.617492  e
+   9 │ 0.05079        9  0.481531  f
 
 julia> f = apply_schema(f, schema(f, df))
 FormulaTerm
@@ -83,15 +83,15 @@ julia> resp, pred = modelcols(f, df);
 
 julia> pred
 9×7 Array{Float64,2}:
- 1.0  1.0  0.986666   0.0  0.0  0.0       0.0
- 1.0  2.0  0.555751   1.0  0.0  0.555751  0.0
- 1.0  3.0  0.437108   0.0  1.0  0.0       0.437108
- 1.0  4.0  0.424718   0.0  0.0  0.0       0.0
- 1.0  5.0  0.773223   1.0  0.0  0.773223  0.0
- 1.0  6.0  0.28119    0.0  1.0  0.0       0.28119
- 1.0  7.0  0.209472   0.0  0.0  0.0       0.0
- 1.0  8.0  0.251379   1.0  0.0  0.251379  0.0
- 1.0  9.0  0.0203749  0.0  1.0  0.0       0.0203749
+ 1.0  1.0  0.236782  0.0  0.0  0.0       0.0
+ 1.0  2.0  0.943741  1.0  0.0  0.943741  0.0
+ 1.0  3.0  0.445671  0.0  1.0  0.0       0.445671
+ 1.0  4.0  0.763679  0.0  0.0  0.0       0.0
+ 1.0  5.0  0.145071  1.0  0.0  0.145071  0.0
+ 1.0  6.0  0.021124  0.0  1.0  0.0       0.021124
+ 1.0  7.0  0.152545  0.0  0.0  0.0       0.0
+ 1.0  8.0  0.617492  1.0  0.0  0.617492  0.0
+ 1.0  9.0  0.481531  0.0  1.0  0.0       0.481531
 
 julia> coefnames(f)
 ("y", ["(Intercept)", "a", "b", "c: e", "c: f", "b & c: e", "b & c: f"])
@@ -224,13 +224,13 @@ StatsModels.TableRegressionModel{LinearModel{GLM.LmResp{Array{Float64,1}},GLM.De
 :(log(y)) ~ 1 + a + b
 
 Coefficients:
-──────────────────────────────────────────────────────────────────────────────
-              Estimate  Std. Error    t value  Pr(>|t|)   Lower 95%  Upper 95%
-──────────────────────────────────────────────────────────────────────────────
-(Intercept)  -4.16168     2.98788   -1.39285     0.2131  -11.4727      3.14939
-a             0.357482    0.342126   1.04489     0.3363   -0.479669    1.19463
-b             2.32528     3.13735    0.741159    0.4866   -5.35154    10.0021
-──────────────────────────────────────────────────────────────────────────────
+──────────────────────────────────────────────────────────────────────────
+                  Coef.  Std. Error      t  Pr(>|t|)  Lower 95%  Upper 95%
+──────────────────────────────────────────────────────────────────────────
+(Intercept)   0.0698025    0.928295   0.08    0.9425  -2.20165    2.34126
+a            -0.105669     0.128107  -0.82    0.4410  -0.419136   0.207797
+b            -1.63199      1.12678   -1.45    0.1977  -4.38911    1.12513
+──────────────────────────────────────────────────────────────────────────
 
 julia> df.log_y = log.(df.y);
 
@@ -240,13 +240,13 @@ StatsModels.TableRegressionModel{LinearModel{GLM.LmResp{Array{Float64,1}},GLM.De
 log_y ~ 1 + a + b
 
 Coefficients:
-──────────────────────────────────────────────────────────────────────────────
-              Estimate  Std. Error    t value  Pr(>|t|)   Lower 95%  Upper 95%
-──────────────────────────────────────────────────────────────────────────────
-(Intercept)  -4.16168     2.98788   -1.39285     0.2131  -11.4727      3.14939
-a             0.357482    0.342126   1.04489     0.3363   -0.479669    1.19463
-b             2.32528     3.13735    0.741159    0.4866   -5.35154    10.0021
-──────────────────────────────────────────────────────────────────────────────
+──────────────────────────────────────────────────────────────────────────
+                  Coef.  Std. Error      t  Pr(>|t|)  Lower 95%  Upper 95%
+──────────────────────────────────────────────────────────────────────────
+(Intercept)   0.0698025    0.928295   0.08    0.9425  -2.20165    2.34126
+a            -0.105669     0.128107  -0.82    0.4410  -0.419136   0.207797
+b            -1.63199      1.12678   -1.45    0.1977  -4.38911    1.12513
+──────────────────────────────────────────────────────────────────────────
 
 ```
 
@@ -256,15 +256,15 @@ interpretation of `+`, `*`, and `&`:
 ```jldoctest 1
 julia> modelmatrix(@formula(y ~ 1 + b + identity(1+b)), df)
 9×3 Array{Float64,2}:
- 1.0  0.986666   1.98667
- 1.0  0.555751   1.55575
- 1.0  0.437108   1.43711
- 1.0  0.424718   1.42472
- 1.0  0.773223   1.77322
- 1.0  0.28119    1.28119
- 1.0  0.209472   1.20947
- 1.0  0.251379   1.25138
- 1.0  0.0203749  1.02037
+ 1.0  0.236782  1.23678
+ 1.0  0.943741  1.94374
+ 1.0  0.445671  1.44567
+ 1.0  0.763679  1.76368
+ 1.0  0.145071  1.14507
+ 1.0  0.021124  1.02112
+ 1.0  0.152545  1.15255
+ 1.0  0.617492  1.61749
+ 1.0  0.481531  1.48153
 ```
 
 ## Constructing a formula programmatically
@@ -350,13 +350,15 @@ simulated coefficients.
 ```jldoctest
 julia> using GLM, DataFrames, StatsModels
 
-julia> data = DataFrame(a = rand(100), b = repeat(["d", "e", "f", "g"], 25));
+julia> using StableRNGs; rng = StableRNG(1);
+
+julia> data = DataFrame(a = rand(rng, 100), b = repeat(["d", "e", "f", "g"], 25));
 
 julia> X = StatsModels.modelmatrix(@formula(y ~ 1 + a*b).rhs, data);
 
 julia> β_true = 1:8;
 
-julia> ϵ = randn(100)*0.1;
+julia> ϵ = randn(rng, 100)*0.1;
 
 julia> data.y = X*β_true .+ ϵ;
 
@@ -366,18 +368,18 @@ StatsModels.TableRegressionModel{LinearModel{GLM.LmResp{Array{Float64,1}},GLM.De
 y ~ 1 + a + b + a & b
 
 Coefficients:
-──────────────────────────────────────────────────────────────────────────
-             Estimate  Std. Error  t value  Pr(>|t|)  Lower 95%  Upper 95%
-──────────────────────────────────────────────────────────────────────────
-(Intercept)   0.98878   0.0384341  25.7266    <1e-43   0.912447    1.06511
-a             2.00843   0.0779388  25.7694    <1e-43   1.85364     2.16323
-b: e          3.03726   0.0616371  49.2764    <1e-67   2.91484     3.15967
-b: f          4.03909   0.0572857  70.5078    <1e-81   3.92531     4.15286
-b: g          5.02948   0.0587224  85.6484    <1e-88   4.91285     5.14611
-a & b: e      5.9385    0.10753    55.2264    <1e-71   5.72494     6.15207
-a & b: f      6.9073    0.112483   61.4075    <1e-75   6.6839      7.1307
-a & b: g      7.93918   0.111285   71.3407    <1e-81   7.71816     8.16021
-──────────────────────────────────────────────────────────────────────────
+───────────────────────────────────────────────────────────────────────
+               Coef.  Std. Error      t  Pr(>|t|)  Lower 95%  Upper 95%
+───────────────────────────────────────────────────────────────────────
+(Intercept)  1.01518   0.0400546  25.34    <1e-42   0.935626    1.09473
+a            1.97476   0.0701427  28.15    <1e-46   1.83545     2.11407
+b: e         3.01269   0.0571186  52.74    <1e-69   2.89925     3.12614
+b: f         4.01918   0.065827   61.06    <1e-75   3.88844     4.14992
+b: g         4.99176   0.0593715  84.08    <1e-88   4.87385     5.10968
+a & b: e     5.98288   0.0954641  62.67    <1e-76   5.79328     6.17248
+a & b: f     6.98622   0.107871   64.76    <1e-77   6.77197     7.20046
+a & b: g     7.92541   0.109873   72.13    <1e-82   7.70719     8.14362
+───────────────────────────────────────────────────────────────────────
 
 ```