restrict distributions' version before its Ext #8

.github/workflows/TagBot.yml

@@ -0,0 +1,35 @@
+
+name: TagBot
+on:
+  issue_comment:
+    types:
+      - created
+  workflow_dispatch:
+    inputs:
+      lookback:
+        default: 3
+permissions:
+  actions: read
+  checks: read
+  contents: write
+  deployments: read
+  issues: read
+  discussions: read
+  packages: read
+  pages: read
+  pull-requests: read
+  repository-projects: read
+  security-events: read
+  statuses: read
+jobs:
+  TagBot:
+    if: github.event_name == 'workflow_dispatch' || github.actor == 'JuliaTagBot'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: JuliaRegistries/TagBot@v1
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          user: szcf-weiya
+          email: szcfweiya@gmail.com
+          # Edit the following line to reflect the actual name of the GitHub Secret containing your private key
+          ssh: ${{ secrets.DOCUMENTER_KEY }}

Project.toml

@@ -29,7 +29,7 @@ StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 StatsPlots = "f3b207a7-027a-5e70-b257-86293d7955fd"
 
 [compat]
-Distributions = "0.25"
+Distributions = "0.25.0 - 0.25.102"
 ECOS = "1.1"
 GLM = "1.8"
 GaussianProcesses = "0.12"

src/mono_decomp.jl

@@ -153,11 +153,6 @@ Perform smoothing spline on `(x, y)`, and make predictions on `xnew`.
 Returns: `yhat`, `ynewhat`,....
 """
 function smooth_spline(x::AbstractVector{T}, y::AbstractVector{T}, xnew::AbstractVector{T}; keep_stuff = false, design_matrix = false, LOOCV = false) where T <: AbstractFloat
-    # if LOOCV
-    #     spl = R"smooth.spline($x, $y, keep.stuff = $keep_stuff, cv = TRUE)"
-    # else
-    #     spl = R"smooth.spline($x, $y, keep.stuff = $keep_stuff, cv = FALSE)"
-    # end
     spl = R"smooth.spline($x, $y, keep.stuff = $keep_stuff, cv = $LOOCV)"
     Σ = nothing
     if keep_stuff
@@ -182,8 +177,8 @@ function cv_smooth_spline(x::AbstractVector{T}, y::AbstractVector{T}, λs::Abstr
     for k in 1:nfold
         test_idx = folds[k]
         train_idx = setdiff(1:n, test_idx)
-        for (i, λ) in enumerate(λs)
-            spl = R"smooth.spline($(x[train_idx]), $(y[train_idx]), lambda = $λ)"
+        for (i, lam) in enumerate(λs)
+            spl = R"smooth.spline($(x[train_idx]), $(y[train_idx]), lambda = $lam)"
             yhat = rcopy(R"predict($spl, $(x[test_idx]))$y")
             # for LOOCV, yhat is a scalar, while y[test_idx] is a vector, use `.-` can help
             errs[k, i] = norm(yhat .- y[test_idx])^2 / length(test_idx) 
@@ -201,11 +196,11 @@ function cv_smooth_spline(x::AbstractVector{T}, y::AbstractVector{T}, λs::Abstr
     elseif ind == 1
         @warn "the optimal is on the left boundary of λs"
     end
-    λopt = λs[ind]
+    optlam = λs[ind]
     if !isnothing(figname)
         savefig(cvplot(μerr, σerr, λs, nfold = nfold, ind0 = ind, lbl = "λ"), figname)
     end
-    spl = R"smooth.spline($x, $y, lambda = $λopt, keep.stuff = TRUE)"
+    spl = R"smooth.spline($x, $y, lambda = $optlam, keep.stuff = TRUE)"
     Σ = recover(rcopy(R"$spl$auxM$Sigma"))
     knots = rcopy(R"$spl$fit$knot")[4:end-3]
     bbasis = R"fda::create.bspline.basis(breaks = $knots, norder = 4)"