Rename dataT to transposed

src/imputors/knn.jl

@@ -45,30 +45,30 @@ function impute!(data::AbstractMatrix{<:Union{T, Missing}},
         impute!(data, Fill(; value=mean, context=ctx))
 
         # transpose to D x N for KDTree
-        dataT = float.(collect(transpose(data)))
-        kdtree = KDTree(dataT, imp.dist)
+        transposed = float.(collect(transpose(data)))
+        kdtree = KDTree(transposed, imp.dist)
 
         # index of columns
-        idxs, dists = NearestNeighbors.knn(kdtree, dataT, imp.num_nn, true)
-        invWdist(i) = dists[i] == 0 ? dataT[idxs[i]] : dists[i]
+        idxs, dists = NearestNeighbors.knn(kdtree, transposed, imp.num_nn, true)
+        invWdist(i) = dists[i] == 0 ? transposed[idxs[i]] : dists[i]
         # TODO : going to parallel?
         for (i, x) in enumerate(mdata)
-            if ndims(dataT) != 1
+            if ndims(transposed) != 1
                 # ndims(dataT) == 1 means there is there is only single row.
                 if mmask[i] == 1
                     # idxs[i] = k-length column indicies
                     # weighted sum
 
                     # what about mulitple missings??
-                    coli = div(i, size(dataT, 1)) + 1
+                    coli = div(i, size(transposed, 1)) + 1
                     # println("$i $coli $(size(mdata))")
                     w = sum(dists[coli]) ./ dists[coli]
                     # Inverse distance weighting
                     if isnan(sum(w)) || isinf(sum(w))
                         # if some distance is zero, keep original value
                         data[i] = data[i]
                     else
-                        data[i] = sum(w .* dataT[idxs[coli]])
+                        data[i] = sum(w .* transposed[idxs[coli]])
                     end
                 end
             end