Fixes #440

Author: cnuernber

CHANGELOG.md

@@ -1,4 +1,9 @@
 # Changelog
+# 7.040
+ * Fix for [issue 450](https://github.com/techascent/tech.ml.dataset/issues/450) - emapped columns could reduce as
+   a different type than declared in the emap declaration. 
+ * Small perf improvements for unique-by.
+ 
 # 7.039
  * Fix error in dtype-next/native-buffer/native-buffer->byte-array
 

deps.edn

@@ -1,6 +1,6 @@
 {:paths ["src" "resources" "target/classes"]
  :deps {;;org.clojure/clojure {:mvn/version "1.11.1"}
-        cnuernber/dtype-next {:mvn/version "10.128"}
+        cnuernber/dtype-next {:mvn/version "10.129"}
         techascent/tech.io {:mvn/version "4.31"
                             :exclusions [org.apache.commons/commons-compress]}
         org.apache.datasketches/datasketches-java {:mvn/version "4.2.0"}

test/tech/v3/dataset_test.clj

@@ -1775,6 +1775,15 @@
              (ds/column :a)
              (vec)))))
 
+(deftest issue-450-incorrect-distinct
+    (is (= 2
+           (->
+            (ds/->dataset {:y [:a :b :b :a :a :a :b :b]})
+            (ds/categorical->number [:y] [] :float64)
+            :y
+            distinct
+            count))))
+
 (comment
   (require '[criterium.core :as crit])
   (def data (vec (repeatedly 100000 (fn [] {:a (rand-int 20) :b (rand) :c (rand)}))))