MeanLabel uses the most frequent y value in classification tasks

src/program/operator.h

@@ -331,7 +331,24 @@ struct Operator<NodeType::MeanLabel, S, Fit>
     using W = typename S::WeightType; 
 
     RetType fit(const Dataset& d, TreeNode& tn) const {
-        tn.data.W = d.y.mean();
+        // we take the mode of the labels if it is a classification problem
+        if (d.classification)
+        {
+            std::unordered_map<float, int> counters;
+            for (float val : d.y) {
+                ++counters[val];
+            }
+
+            auto mode = std::max_element(
+                counters.begin(), counters.end(),
+                [](const auto& a, const auto& b) { return a.second < b.second; }
+            );
+
+            tn.data.W = mode->first;
+        }
+        else
+            tn.data.W = d.y.mean();
+
         return predict(d, tn);
     };