PR corrections

Author: wildart

doc/source/validate.md

@@ -103,8 +103,8 @@ mutualinfo
 ## Confusion matrix
 
 Pair [confusion matrix](https://en.wikipedia.org/wiki/Confusion_matrix)
-arising from two clusterings is a 2x2 contingency table representation of
-the partition co-occurrence table, see [`counts`](@ref).
+arising from two clusterings is a 2×2 contingency table representation of
+the partition co-occurrence, see [`counts`](@ref).
 
 ```@docs
 confusion

src/confusion.jl

@@ -1,23 +1,21 @@
 """
-    confusion(a::ClusteringResult, b::ClusteringResult) -> Matrix{Int}
-    confusion(a::ClusteringResult, b::AbstractVector{<:Integer}) -> Matrix{Int}
-    confusion(a::AbstractVector{<:Integer}, b::ClusteringResult) -> Matrix{Int}
-    confusion(a::AbstractVector{<:Integer}, b::AbstractVector{<:Integer}) -> Matrix{Int}
+    confusion(a::Union{ClusteringResult, AbstractVector},
+              b::Union{ClusteringResult, AbstractVector}) -> Matrix{Int}
 
-Return 2x2 confusion matrix `C` that represents partition co-occurrence or
+Return 2×2 confusion matrix `C` that represents partition co-occurrence or
 similarity matrix between two clusterings by considering all pairs of samples
 and counting  pairs that are assigned into the same or into different clusters
 under the true and predicted clusterings.
 
 Considering a pair of samples that is in the same group as a **positive pair**,
 and a pair is in the different group as a **negative pair**, then the count of
-true positives is `C₀₀`, false negatives is `C₀₁`, false positives `C₁₀`, and
-true negatives is `C₁₁`:
+true positives is `C₁₁`, false negatives is `C₁₂`, false positives `C₂₁`, and
+true negatives is `C₂₂`:
 
 |  | Positive | Negative |
 |:--:|:-:|:-:|
-|Positive|C₀₀|C₁₀|
-|Negative|C₀₁|C₁₁|
+|Positive|C₁₁|C₁₂|
+|Negative|C₂₁|C₂₂|
 """
 function confusion(a::AbstractVector{<:Integer}, b::AbstractVector{<:Integer})
     c = counts(a, b)
@@ -27,10 +25,11 @@ function confusion(a::AbstractVector{<:Integer}, b::AbstractVector{<:Integer})
     njs = sum(abs2, sum(c, dims=1))        # sum of squares of sums of columns
 
     t2 = sum(abs2, c)                      # sum over rows & columns of nij^2
-    t3 = nis+njs
-    C = Int[(t2-n)/2 (nis-t2)/2; (njs-t2)/2 (t2+n^2-t3)/2]
+    t3 = nis + njs
+    C = [(t2 - n)÷2 (nis - t2)÷2; (njs - t2)÷2 (t2 + n^2 - t3)÷2]
     return C
 end
+
 confusion(a::ClusteringResult, b::ClusteringResult) =
     confusion(assignments(a), assignments(b))
 confusion(a::AbstractVector{<:Integer}, b::ClusteringResult) =

src/randindex.jl

@@ -14,23 +14,23 @@ Returns a tuple of indices:
 
 # References
 > Lawrence Hubert and Phipps Arabie (1985). *Comparing partitions.*
-> Journal of Classification 2 (1): 193–218
+> Journal of Classification 2 (1): 193-218
 
 > Meila, Marina (2003). *Comparing Clusterings by the Variation of
-> Information.* Learning Theory and Kernel Machines: 173–187.
+> Information.* Learning Theory and Kernel Machines: 173-187.
 
-> Steinley, Douglas (2004). *Properties of the Hubert–Arabie Adjusted
+> Steinley, Douglas (2004). *Properties of the Hubert-Arabie Adjusted
 > Rand Index.* Psychological Methods, Vol. 9, No. 3: 386-396
 """
 function randindex(a, b)
-    a, c, b, d = confusion(a,b) # Table 2 from Steinley 2004
+    c11, c21, c12, c22 = confusion(a, b) # Table 2 from Steinley 2004
 
-    t = a+ b + c + d   # total number of pairs of entities
-    A = a + d
-    D = b + c
+    t = c11 + c12 + c21 + c22   # total number of pairs of entities
+    A = c11 + c22
+    D = c12 + c21
 
     # expected index
-    ERI = (a+b)*(a+c)+(c+d)*(b+d)
+    ERI = (c11+c12)*(c11+c21)+(c21+c22)*(c12+c22)
     # adjusted Rand - Hubert & Arabie 1985
     ARI = D == 0 ? 1.0 : (t*A-ERI)/(t*t-ERI) # (9) from Steinley 2004