Add ML Predict, Train APIs (#755)

* Added ML predict and train APIs, updated CHANGELOG. Signed-off-by: Nathalie Jonathan <[email protected]> * Added ML train and predict API, fixed test for predict and train API, defined ML status as an enum, updated ML task state enum, updated CHANGELOG.md. Signed-off-by: Nathalie Jonathan <[email protected]> * Updated examples to use books themed data, removed output from predict and train_predict API, fixed vale complaints. Signed-off-by: Nathalie Jonathan <[email protected]> * Removed excluded parts, changed files path. Signed-off-by: Nathalie Jonathan <[email protected]> --------- Signed-off-by: Nathalie Jonathan <[email protected]>
opensearch-project · Dec 31, 2024 · a3849b4 · a3849b4
1 parent 2c6502d
commit a3849b4
Show file tree

Hide file tree

Showing 6 changed files with 565 additions and 0 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -40,6 +40,7 @@ Inspired from [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 - Added support for combining output variables ([#737](https://github.com/opensearch-project/opensearch-api-specification/pull/737))
 - Added 404 response to `/_search/scroll` ([#749](https://github.com/opensearch-project/opensearch-api-specification/pull/749))
 - Added `node_failures` to `DELETE /_search/scroll` and `DELETE /_search/scroll/{scroll_id}` ([#749](https://github.com/opensearch-project/opensearch-api-specification/pull/749))
+- Added `POST /_plugins/_ml/_train/{algorithm_name}`, `_predict/{algorithm_name}/{model_id}`, and `_train_predict/{algorithm_name}` ([#755](https://github.com/opensearch-project/opensearch-api-specification/pull/755))
 
 ### Removed
 - Removed unsupported `_common.mapping:SourceField`'s `mode` field and associated `_common.mapping:SourceFieldMode` enum ([#652](https://github.com/opensearch-project/opensearch-api-specification/pull/652))

diff --git a/spec/namespaces/ml.yaml b/spec/namespaces/ml.yaml
@@ -102,6 +102,43 @@ paths:
       responses:
         '200':
           $ref: '#/components/responses/ml.search_models@200'
+  /_plugins/_ml/_predict/{algorithm_name}/{model_id}:
+    post:
+      operationId: ml.predict.0
+      x-operation-group: ml.predict
+      description: Predicts new data with trained model.
+      parameters:
+        - $ref: '#/components/parameters/ml.predict::path.algorithm_name'
+        - $ref: '#/components/parameters/ml.predict::path.model_id'
+      requestBody:
+        $ref: '#/components/requestBodies/ml.predict'
+      responses:
+        '200':
+          $ref: '#/components/responses/ml.predict@200'
+  /_plugins/_ml/_train/{algorithm_name}:
+    post:
+      operationId: ml.train.0
+      x-operation-group: ml.train
+      description: Trains a model synchronously.
+      parameters:
+        - $ref: '#/components/parameters/ml.train::path.algorithm_name'
+      requestBody:
+        $ref: '#/components/requestBodies/ml.train'
+      responses:
+        '200':
+          $ref: '#/components/responses/ml.train@200'
+  /_plugins/_ml/_train_predict/{algorithm_name}:
+    post:
+      operationId: ml.train_predict.0
+      x-operation-group: ml.train_predict
+      description: Trains a model and predicts against the same training dataset.
+      parameters:
+        - $ref: '#/components/parameters/ml.train_predict::path.algorithm_name'
+      requestBody:
+        $ref: '#/components/requestBodies/ml.train_predict'
+      responses:
+        '200':
+          $ref: '#/components/responses/ml.train_predict@200'
   /_plugins/_ml/connectors/_create:
     post:
       operationId: ml.create_connector.0
@@ -212,6 +249,64 @@ components:
             required:
               - query
               - size
+    ml.predict:
+      content:
+        application/json:
+          schema:
+            type: object
+            properties:
+              input_query:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/InputQuery'
+              input_index:
+                type: array
+                items:
+                  type: string
+                description: The input index.
+              text_docs:
+                type: array
+                items:
+                  type: string
+                description: The text documents.
+              return_number:
+                type: boolean
+                description: Whether to return bytes in model output.
+              target_response:
+                type: array
+                items:
+                  type: string
+                description: The target response.
+    ml.train:
+      content:
+        application/json:
+          schema:
+            type: object
+            properties:
+              parameters:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/TrainParameters'
+              input_query:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/InputQuery'
+              input_index:
+                type: array
+                description: The input index.
+                items:
+                  type: string
+    ml.train_predict:
+      content:
+        application/json:
+          schema:
+            type: object
+            properties:
+              parameters:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/TrainParameters'
+              input_query:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/InputQuery'
+              input_index:
+                type: array
+                description: The input index.
+                items:
+                  type: string
+              input_data:
+                $ref: '../schemas/ml._common.yaml#/components/schemas/PredictionResult'
     ml.create_connector:
       content:
         application/json:
@@ -340,6 +435,21 @@ components:
         application/json:
           schema:
             $ref: '../schemas/ml._common.yaml#/components/schemas/SearchModelsResponse'
+    ml.predict@200:
+      content:
+        application/json:
+          schema:
+            $ref: '../schemas/ml._common.yaml#/components/schemas/PredictResponse'
+    ml.train@200:
+      content:
+        application/json:
+          schema:
+            $ref: '../schemas/ml._common.yaml#/components/schemas/TrainResponse'
+    ml.train_predict@200:
+      content:
+        application/json:
+          schema:
+            $ref: '../schemas/ml._common.yaml#/components/schemas/TrainPredictResponse'
     ml.create_connector@200:
       content:
         application/json:
@@ -403,6 +513,30 @@ components:
       required: true
       schema:
         type: string
+    ml.predict::path.algorithm_name:
+      name: algorithm_name
+      in: path
+      required: true
+      schema:
+        type: string
+    ml.predict::path.model_id:
+      name: model_id
+      in: path
+      required: true
+      schema:
+        type: string
+    ml.train::path.algorithm_name:
+      name: algorithm_name
+      in: path
+      required: true
+      schema:
+        type: string
+    ml.train_predict::path.algorithm_name:
+      name: algorithm_name
+      in: path
+      required: true
+      schema:
+        type: string
     ml.delete_connector::path.connector_id:
       name: connector_id
       in: path