[parser] for the min updates method to work, the feature that have not yet reached the min_updates threshold also need to be ignored when scoring, that way the model has to perform without those features, and should make more updates if they're relevant

src/averaged_perceptron_trainer.c

@@ -316,20 +316,27 @@ uint32_t averaged_perceptron_trainer_predict(averaged_perceptron_trainer_t *self
 
     double_array_zero(scores->a, scores->n);
 
+    uint64_t *update_counts = self->update_counts->a;
+
     cstring_array_foreach(features, i, feature, {
         if (!averaged_perceptron_trainer_get_feature_id(self, feature, &feature_id, add_if_missing)) {
             continue;
         }
 
-        weights = averaged_perceptron_trainer_get_class_weights(self, feature_id, add_if_missing);
+        uint64_t update_count = update_counts[feature_id];
+        bool keep_feature = update_count >= self->min_updates;
 
-        if (weights == NULL) {
-            continue;
+        if (keep_feature) {
+            weights = averaged_perceptron_trainer_get_class_weights(self, feature_id, add_if_missing);
+
+            if (weights == NULL) {
+                continue;
+            }
+
+            kh_foreach(weights, class_id, weight, {
+                scores->a[class_id] += weight.value;
+            })
         }
-
-        kh_foreach(weights, class_id, weight, {
-            scores->a[class_id] += weight.value;
-        })
     })
 
     int64_t max_score = double_array_argmax(scores->a, scores->n);