Reuse mask in each thread

src/forest.jl

@@ -226,8 +226,9 @@ function _view_X_y!(mask, X, y, splitpoint::SplitPoint, comparison)
         result = comparison(value, splitpoint.value)
         @inbounds mask[i] = result
     end
-    X_view = @inbounds view(X, mask, :)
-    y_view = @inbounds view(y, mask)
+    mask_subset = view(mask, 1:length(y))
+    X_view = @inbounds view(X, mask_subset, :)
+    y_view = @inbounds view(y, mask_subset)
     return (X_view, y_view)
 end
 
@@ -262,8 +263,9 @@ Arguments:
     During random forest creation, the number of split candidates is limited to make the trees less correlated.
     See Section 8.2.2 of https://doi.org/10.1007/978-1-0716-1418-1 for details.
 """
-function _tree(
+function _tree!(
         rng::AbstractRNG,
+        mask::Vector{Bool},
         X,
         y::AbstractVector,
         classes::AbstractVector,
@@ -288,14 +290,13 @@ function _tree(
     end
     depth += 1
 
-    mask = Vector{Bool}(undef, length(y))
     left = let
         _X, yl = _view_X_y!(mask, X, y, sp, <)
-        _tree(rng, _X, yl, classes, colnames; cutpoints, depth, max_depth)
+        _tree!(rng, mask, _X, yl, classes, colnames; cutpoints, depth, max_depth)
     end
     right = let
         _X, yr = _view_X_y!(mask, X, y, sp, ≥)
-        _tree(rng, _X, yr, classes, colnames; cutpoints, depth, max_depth)
+        _tree!(rng, mask, _X, yr, classes, colnames; cutpoints, depth, max_depth)
     end
     node = Node(sp, left, right)
     return node
@@ -391,8 +392,10 @@ function _forest(
         rows = rand(_rng, 1:length(y), n_samples)
         _X = view(X, rows, :)
         _y = view(y, rows)
-        tree = _tree(
+        mask = Vector{Bool}(undef, length(y))
+        tree = _tree!(
             _rng,
+            mask,
             _X,
             _y,
             classes,