[ML] Support the unsigned_long type in data frame analytics (#64066)

Adds support for the unsigned_long type to data frame analytics.

This type is handled in the same way as the long type.  Values
sent to the ML native processes are converted to floats and
hence will lose accuracy when outside the range where a float
can uniquely represent long values.

Relates #60050

x-pack/plugin/core/src/main/java/org/elasticsearch/xpack/core/ml/dataframe/analyses/Types.java

@@ -28,12 +28,13 @@ private Types() {}
             .collect(Collectors.toUnmodifiableSet());
 
     private static final Set<String> NUMERICAL_TYPES =
-        Stream.concat(Stream.of(NumberType.values()).map(NumberType::typeName), Stream.of("scaled_float"))
+        Stream.concat(Stream.of(NumberType.values()).map(NumberType::typeName), Stream.of("scaled_float", "unsigned_long"))
             .collect(Collectors.toUnmodifiableSet());
 
     private static final Set<String> DISCRETE_NUMERICAL_TYPES =
-        Stream.of(NumberType.BYTE, NumberType.SHORT, NumberType.INTEGER, NumberType.LONG)
-            .map(NumberType::typeName)
+        Stream.concat(
+            Stream.of(NumberType.BYTE, NumberType.SHORT, NumberType.INTEGER, NumberType.LONG).map(NumberType::typeName),
+            Stream.of("unsigned_long"))
             .collect(Collectors.toUnmodifiableSet());
 
     private static final Set<String> BOOL_TYPES = Collections.singleton(BooleanFieldMapper.CONTENT_TYPE);

x-pack/plugin/ml/qa/native-multi-node-tests/src/javaRestTest/java/org/elasticsearch/xpack/ml/integration/DataFrameAnalysisCustomFeatureIT.java

@@ -167,7 +167,7 @@ private static void createIndex(String index, boolean isDatastream) {
             "          \"type\": \"double\"\n" +
             "        }," +
             "        \""+ DISCRETE_NUMERICAL_FIELD + "\": {\n" +
-            "          \"type\": \"integer\"\n" +
+            "          \"type\": \"unsigned_long\"\n" +
             "        }," +
             "        \""+ TEXT_FIELD + "\": {\n" +
             "          \"type\": \"text\"\n" +

x-pack/plugin/ml/qa/native-multi-node-tests/src/javaRestTest/java/org/elasticsearch/xpack/ml/integration/ExplainDataFrameAnalyticsIT.java

@@ -52,7 +52,7 @@ public void testSourceQueryIsApplied() throws IOException {
         client().admin().indices().prepareCreate(sourceIndex)
             .setMapping(
                 "numeric_1", "type=double",
-                "numeric_2", "type=float",
+                "numeric_2", "type=unsigned_long",
                 "categorical", "type=keyword",
                 "filtered_field", "type=keyword")
             .get();
@@ -64,7 +64,7 @@ public void testSourceQueryIsApplied() throws IOException {
             IndexRequest indexRequest = new IndexRequest(sourceIndex);
             indexRequest.source(
                 "numeric_1", 1.0,
-                "numeric_2", 2.0,
+                "numeric_2", 2,
                 "categorical", i % 2 == 0 ? "class_1" : "class_2",
                 "filtered_field", i < 2 ? "bingo" : "rest"); // We tag bingo on the first two docs to ensure we have 2 classes
             bulkRequestBuilder.add(indexRequest);

x-pack/plugin/ml/qa/native-multi-node-tests/src/javaRestTest/java/org/elasticsearch/xpack/ml/integration/RegressionIT.java

@@ -698,7 +698,7 @@ static void indexData(String sourceIndex, int numTrainingRows, int numNonTrainin
             "          \"type\": \"double\"\n" +
             "        }," +
             "        \"" + DISCRETE_NUMERICAL_FEATURE_FIELD + "\": {\n" +
-            "          \"type\": \"long\"\n" +
+            "          \"type\": \"unsigned_long\"\n" +
             "        }," +
             "        \"" + DEPENDENT_VARIABLE_FIELD + "\": {\n" +
             "          \"type\": \"double\"\n" +

x-pack/plugin/ml/qa/native-multi-node-tests/src/javaRestTest/java/org/elasticsearch/xpack/ml/integration/RunDataFrameAnalyticsIT.java

@@ -71,7 +71,7 @@ public void testOutlierDetectionWithFewDocuments() throws Exception {
         String sourceIndex = "test-outlier-detection-with-few-docs";
 
         client().admin().indices().prepareCreate(sourceIndex)
-            .setMapping("numeric_1", "type=double", "numeric_2", "type=float", "categorical_1", "type=keyword")
+            .setMapping("numeric_1", "type=double", "numeric_2", "type=unsigned_long", "categorical_1", "type=keyword")
             .get();
 
         BulkRequestBuilder bulkRequestBuilder = client().prepareBulk();
@@ -83,7 +83,7 @@ public void testOutlierDetectionWithFewDocuments() throws Exception {
             // We insert one odd value out of 5 for one feature
             String docId = i == 0 ? "outlier" : "normal" + i;
             indexRequest.id(docId);
-            indexRequest.source("numeric_1", i == 0 ? 100.0 : 1.0, "numeric_2", 1.0, "categorical_1", "foo_" + i);
+            indexRequest.source("numeric_1", i == 0 ? 100.0 : 1.0, "numeric_2", 1, "categorical_1", "foo_" + i);
             bulkRequestBuilder.add(indexRequest);
         }
         BulkResponse bulkResponse = bulkRequestBuilder.get();

x-pack/plugin/ml/src/test/java/org/elasticsearch/xpack/ml/dataframe/extractor/ExtractedFieldsDetectorTests.java

@@ -105,7 +105,7 @@ public void testDetect_GivenOutlierDetectionAndNonNumericField() {
         assertThat(fieldExtraction.v2().get(0).getName(), equalTo("some_keyword"));
         assertThat(fieldExtraction.v2().get(0).isIncluded(), is(false));
         assertThat(fieldExtraction.v2().get(0).getReason(), equalTo("unsupported type; supported types are " +
-            "[boolean, byte, double, float, half_float, integer, long, scaled_float, short]"));
+            "[boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"));
     }
 
     public void testDetect_GivenOutlierDetectionAndFieldWithNumericAndNonNumericTypes() {
@@ -121,7 +121,7 @@ public void testDetect_GivenOutlierDetectionAndFieldWithNumericAndNonNumericType
         assertThat(fieldExtraction.v2().get(0).getName(), equalTo("indecisive_field"));
         assertThat(fieldExtraction.v2().get(0).isIncluded(), is(false));
         assertThat(fieldExtraction.v2().get(0).getReason(), equalTo("unsupported type; supported types are " +
-            "[boolean, byte, double, float, half_float, integer, long, scaled_float, short]"));
+            "[boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"));
     }
 
     public void testDetect_GivenOutlierDetectionAndMultipleFields() {
@@ -147,7 +147,7 @@ public void testDetect_GivenOutlierDetectionAndMultipleFields() {
             FieldSelection.included("some_boolean", Collections.singleton("boolean"), false, FieldSelection.FeatureType.NUMERICAL),
             FieldSelection.included("some_float", Collections.singleton("float"), false, FieldSelection.FeatureType.NUMERICAL),
             FieldSelection.excluded("some_keyword", Collections.singleton("keyword"), "unsupported type; " +
-                "supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short]"),
+                "supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"),
             FieldSelection.included("some_long", Collections.singleton("long"), false, FieldSelection.FeatureType.NUMERICAL)
         );
     }
@@ -282,7 +282,7 @@ public void testDetect_GivenRegressionAndRequiredFieldHasInvalidType() {
         ElasticsearchStatusException e = expectThrows(ElasticsearchStatusException.class, extractedFieldsDetector::detect);
 
         assertThat(e.getMessage(), equalTo("invalid types [keyword] for required field [foo]; " +
-            "expected types are [byte, double, float, half_float, integer, long, scaled_float, short]"));
+            "expected types are [byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"));
     }
 
     public void testDetect_GivenClassificationAndRequiredFieldHasInvalidType() {
@@ -298,7 +298,7 @@ public void testDetect_GivenClassificationAndRequiredFieldHasInvalidType() {
         ElasticsearchStatusException e = expectThrows(ElasticsearchStatusException.class, extractedFieldsDetector::detect);
 
         assertThat(e.getMessage(), equalTo("invalid types [float] for required field [some_float]; " +
-            "expected types are [boolean, byte, integer, ip, keyword, long, short, text]"));
+            "expected types are [boolean, byte, integer, ip, keyword, long, short, text, unsigned_long]"));
     }
 
     public void testDetect_GivenClassificationAndDependentVariableHasInvalidCardinality() {
@@ -371,7 +371,8 @@ public void testDetect_GivenExcludedFieldIsUnsupported() {
 
         assertFieldSelectionContains(fieldExtraction.v2(),
             FieldSelection.excluded("categorical", Collections.singleton("keyword"),
-                "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short]"),
+                "unsupported type; supported types are " +
+                    "[boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"),
             FieldSelection.included("numeric", Collections.singleton("float"), false, FieldSelection.FeatureType.NUMERICAL)
         );
     }
@@ -471,7 +472,7 @@ public void testDetect_GivenIncludedFieldHasUnsupportedType() {
         ElasticsearchStatusException e = expectThrows(ElasticsearchStatusException.class, extractedFieldsDetector::detect);
 
         assertThat(e.getMessage(), equalTo("field [your_keyword] has unsupported type [keyword]. " +
-            "Supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short]."));
+            "Supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]."));
     }
 
     public void testDetect_GivenNotIncludedFieldHasUnsupportedType() {
@@ -492,7 +493,8 @@ public void testDetect_GivenNotIncludedFieldHasUnsupportedType() {
 
         assertFieldSelectionContains(fieldExtraction.v2(),
             FieldSelection.excluded("categorical", Collections.singleton("keyword"),
-                "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, long, scaled_float, short]"),
+                "unsupported type; supported types are " +
+                    "[boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]"),
             FieldSelection.included("numeric", Collections.singleton("float"), false, FieldSelection.FeatureType.NUMERICAL)
         );
     }
@@ -517,7 +519,7 @@ public void testDetect_GivenIndexContainsResultsField() {
             FieldSelection.included("my_field1", Collections.singleton("float"), false, FieldSelection.FeatureType.NUMERICAL),
             FieldSelection.included("your_field2", Collections.singleton("float"), false, FieldSelection.FeatureType.NUMERICAL),
             FieldSelection.excluded("your_keyword", Collections.singleton("keyword"), "unsupported type; supported types " +
-                "are [boolean, byte, double, float, half_float, integer, long, scaled_float, short]")
+                "are [boolean, byte, double, float, half_float, integer, long, scaled_float, short, unsigned_long]")
         );
     }
 

x-pack/plugin/src/test/resources/rest-api-spec/test/ml/explain_data_frame_analytics.yml

@@ -225,7 +225,7 @@
   - match: { field_selection.2.is_included: false }
   - match: { field_selection.2.is_required: false }
   - is_false: field_selection.2.feature_type
-  - match: { field_selection.2.reason: "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, ip, keyword, long, scaled_float, short, text]" }
+  - match: { field_selection.2.reason: "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, ip, keyword, long, scaled_float, short, text, unsigned_long]" }
   - match: { field_selection.3.name: "field_4" }
   - match: { field_selection.3.mapping_types: ["text"] }
   - match: { field_selection.3.is_included: false }
@@ -299,7 +299,7 @@
   - match: { field_selection.2.is_included: false }
   - match: { field_selection.2.is_required: false }
   - is_false: field_selection.2.feature_type
-  - match: { field_selection.2.reason: "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, ip, keyword, long, scaled_float, short, text]" }
+  - match: { field_selection.2.reason: "unsupported type; supported types are [boolean, byte, double, float, half_float, integer, ip, keyword, long, scaled_float, short, text, unsigned_long]" }
   - match: { field_selection.3.name: "field_4" }
   - match: { field_selection.3.mapping_types: ["text"] }
   - match: { field_selection.3.is_included: false }