Add XGBoost if-inversion for missing features

4 files changed, 9 insertions, 40 deletions

diff --git a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostParser.java b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostParser.java
index 9de07eed475..c41a114a970 100644
--- a/model-integration/src/main/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostParser.java
+++ b/model-integration/src/main/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostParser.java
@@ -16,7 +16,6 @@ import com.fasterxml.jackson.databind.ObjectMapper;
 class XGBoostParser {
 
     private List<XGBoostTree> xgboostTrees;
-    private boolean doIfInversion = false;
 
     /**
      * Constructor stores parsed JSON trees.
@@ -32,7 +31,6 @@ class XGBoostParser {
         for (JsonNode treeNode : forestNode) {
             this.xgboostTrees.add(mapper.treeToValue(treeNode, XGBoostTree.class));
         }
-        doIfInversion = filePath.endsWith("if_inversion.json");
     }
 
     /**
@@ -71,9 +69,12 @@ class XGBoostParser {
                 trueExp = treeToRankExp(node.getChildren().get(1));
                 falseExp = treeToRankExp(node.getChildren().get(0));
             }
-            String condition = node.getSplit() + " < " + node.getSplit_condition();
-            if (doIfInversion && node.getMissing() == node.getYes()) {
+            String condition;
+            if (node.getMissing() == node.getYes()) {
+                // Note: this is for handling missing features, as the backend handles comparison with NaN as false.
                 condition = "!(" + node.getSplit() + " >= " + node.getSplit_condition() + ")";
+            } else {
+                condition = node.getSplit() + " < " + node.getSplit_condition();
             }
             return "if (" + condition + ", " + trueExp + ", " + falseExp + ")";
         }
diff --git a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostImportTestCase.java b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostImportTestCase.java
index 6d75b9f40ff..9a71905aa65 100644
--- a/model-integration/src/test/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostImportTestCase.java
+++ b/model-integration/src/test/java/ai/vespa/rankingexpression/importer/xgboost/XGBoostImportTestCase.java
@@ -20,22 +20,9 @@ public class XGBoostImportTestCase {
         assertEquals(1, model.expressions().size());
         RankingExpression expression = model.expressions().get("test");
         assertNotNull(expression);
-        assertEquals("if (f29 < -0.1234567, if (f56 < -0.242398, 1.71218, -1.70044), if (f109 < 0.8723473, -1.94071, 1.85965)) + if (f60 < -0.482947, if (f29 < -4.2387498, 0.784718, -0.96853), -6.23624)",
-                     expression.getRoot().toString());
-        assertEquals(1, model.outputExpressions().size());
-    }
-
-    @Test
-    public void testXGBoostWithIfInversion() {
-        ImportedModel model = new XGBoostImporter().importModel("test", "src/test/models/xgboost/xgboost.2.2.if_inversion.json");
-        assertTrue("All inputs are scalar", model.inputs().isEmpty());
-        assertEquals(1, model.expressions().size());
-        RankingExpression expression = model.expressions().get("test");
-        assertNotNull(expression);
-        assertEquals("if (f29 < -0.1234567, if (!(f56 >= -0.242398), 1.71218, -1.70044), if (f109 < 0.8723473, -1.94071, 1.85965)) + if (f60 < -0.482947, if (!(f29 >= -4.2387498), 0.784718, -0.96853), -6.23624)",
+        assertEquals("if (f29 < -0.1234567, if (!(f56 >= -0.242398), 1.71218, -1.70044), if (f109 < 0.8723473, -1.94071, 1.85965)) + if (!(f60 >= -0.482947), if (f29 < -4.2387498, 0.784718, -0.96853), -6.23624)",
                 expression.getRoot().toString());
         assertEquals(1, model.outputExpressions().size());
     }
 
-
 }
diff --git a/model-integration/src/test/models/xgboost/xgboost.2.2.if_inversion.json b/model-integration/src/test/models/xgboost/xgboost.2.2.if_inversion.json
deleted file mode 100644
index c1217ef55a6..00000000000
--- a/model-integration/src/test/models/xgboost/xgboost.2.2.if_inversion.json
+++ /dev/null
@@ -1,19 +0,0 @@
-[
-  { "nodeid": 0, "depth": 0, "split": "f29", "split_condition": -0.1234567, "yes": 1, "no": 2, "missing": 2, "children": [
-    { "nodeid": 1, "depth": 1, "split": "f56", "split_condition": -0.242398, "yes": 3, "no": 4, "missing": 3, "children": [
-      { "nodeid": 3, "leaf": 1.71218 },
-      { "nodeid": 4, "leaf": -1.70044 }
-    ]},
-    { "nodeid": 2, "depth": 1, "split": "f109", "split_condition": 0.8723473, "yes": 5, "no": 6, "missing": 6, "children": [
-      { "nodeid": 5, "leaf": -1.94071 },
-      { "nodeid": 6, "leaf": 1.85965 }
-    ]}
-  ]},
-  { "nodeid": 0, "depth": 0, "split": "f60", "split_condition": -0.482947, "yes": 1, "no": 2, "missing": 2, "children": [
-    { "nodeid": 1, "depth": 1, "split": "f29", "split_condition": -4.2387498, "yes": 3, "no": 4, "missing": 3, "children": [
-      { "nodeid": 3, "leaf": 0.784718 },
-      { "nodeid": 4, "leaf": -0.96853 }
-    ]},
-    { "nodeid": 2, "leaf": -6.23624 }
-  ]}
-]
\ No newline at end of file
diff --git a/model-integration/src/test/models/xgboost/xgboost.2.2.json b/model-integration/src/test/models/xgboost/xgboost.2.2.json
index f8949b47e52..a18b9966b55 100644
--- a/model-integration/src/test/models/xgboost/xgboost.2.2.json
+++ b/model-integration/src/test/models/xgboost/xgboost.2.2.json
@@ -1,16 +1,16 @@
 [
-  { "nodeid": 0, "depth": 0, "split": "f29", "split_condition": -0.1234567, "yes": 1, "no": 2, "missing": 1, "children": [
+  { "nodeid": 0, "depth": 0, "split": "f29", "split_condition": -0.1234567, "yes": 1, "no": 2, "missing": 2, "children": [
     { "nodeid": 1, "depth": 1, "split": "f56", "split_condition": -0.242398, "yes": 3, "no": 4, "missing": 3, "children": [
       { "nodeid": 3, "leaf": 1.71218 },
       { "nodeid": 4, "leaf": -1.70044 }
     ]},
-    { "nodeid": 2, "depth": 1, "split": "f109", "split_condition": 0.8723473, "yes": 5, "no": 6, "missing": 5, "children": [
+    { "nodeid": 2, "depth": 1, "split": "f109", "split_condition": 0.8723473, "yes": 5, "no": 6, "missing": 6, "children": [
       { "nodeid": 5, "leaf": -1.94071 },
       { "nodeid": 6, "leaf": 1.85965 }
     ]}
   ]},
   { "nodeid": 0, "depth": 0, "split": "f60", "split_condition": -0.482947, "yes": 1, "no": 2, "missing": 1, "children": [
-    { "nodeid": 1, "depth": 1, "split": "f29", "split_condition": -4.2387498, "yes": 3, "no": 4, "missing": 3, "children": [
+    { "nodeid": 1, "depth": 1, "split": "f29", "split_condition": -4.2387498, "yes": 3, "no": 4, "missing": 4, "children": [
       { "nodeid": 3, "leaf": 0.784718 },
       { "nodeid": 4, "leaf": -0.96853 }
     ]},