Examples with MLInputDataset org.opensearch.ml.common.dataset.MLInputDataset used on opensource projects

Example 16 with MLInputDataset

use of org.opensearch.ml.common.dataset.MLInputDataset in project ml-commons by opensearch-project.

the class MLEngineTest method trainLinearRegressionModel.

private Model trainLinearRegressionModel() {
    LinearRegressionParams parameters = LinearRegressionParams.builder().objectiveType(LinearRegressionParams.ObjectiveType.SQUARED_LOSS).optimizerType(LinearRegressionParams.OptimizerType.ADAM).learningRate(0.01).epsilon(1e-6).beta1(0.9).beta2(0.99).target("price").build();
    DataFrame trainDataFrame = constructLinearRegressionTrainDataFrame();
    MLInputDataset inputDataset = DataFrameInputDataset.builder().dataFrame(trainDataFrame).build();
    Input mlInput = MLInput.builder().algorithm(FunctionName.LINEAR_REGRESSION).parameters(parameters).inputDataset(inputDataset).build();
    return MLEngine.train(mlInput);
}

Example 17 with MLInputDataset

use of org.opensearch.ml.common.dataset.MLInputDataset in project ml-commons by opensearch-project.

the class MLEngineTest method predictKMeans.

@Test
public void predictKMeans() {
    Model model = trainKMeansModel();
    DataFrame predictionDataFrame = constructKMeansDataFrame(10);
    MLInputDataset inputDataset = DataFrameInputDataset.builder().dataFrame(predictionDataFrame).build();
    Input mlInput = MLInput.builder().algorithm(FunctionName.KMEANS).inputDataset(inputDataset).build();
    MLPredictionOutput output = (MLPredictionOutput) MLEngine.predict(mlInput, model);
    DataFrame predictions = output.getPredictionResult();
    Assert.assertEquals(10, predictions.size());
    predictions.forEach(row -> Assert.assertTrue(row.getValue(0).intValue() == 0 || row.getValue(0).intValue() == 1));
}

Example 18 with MLInputDataset

use of org.opensearch.ml.common.dataset.MLInputDataset in project ml-commons by opensearch-project.

the class RestMLTrainAndPredictIT method trainAndPredictKmeansWithParmas.

private void trainAndPredictKmeansWithParmas(KMeansParams params, Consumer<Map<Double, Integer>> function) throws IOException {
    SearchSourceBuilder sourceBuilder = new SearchSourceBuilder();
    sourceBuilder.query(new MatchAllQueryBuilder());
    sourceBuilder.size(1000);
    sourceBuilder.fetchSource(new String[] { "petal_length_in_cm", "petal_width_in_cm" }, null);
    MLInputDataset inputData = SearchQueryInputDataset.builder().indices(ImmutableList.of(irisIndex)).searchSourceBuilder(sourceBuilder).build();
    trainAndPredictKmeansWithIrisData(params, inputData, function);
}

Example 19 with MLInputDataset

use of org.opensearch.ml.common.dataset.MLInputDataset in project ml-commons by opensearch-project.

the class PredictionITTests method initTestingData.

@Before
public void initTestingData() throws ExecutionException, InterruptedException {
    generateMLTestingData();
    SearchSourceBuilder searchSourceBuilder = generateSearchSourceBuilder();
    MLInputDataset inputDataset = new SearchQueryInputDataset(Collections.singletonList(TESTING_INDEX_NAME), searchSourceBuilder);
    taskId = trainModel(inputDataset);
    waitModelAvailable(taskId);
}

Example 20 with MLInputDataset

use of org.opensearch.ml.common.dataset.MLInputDataset in project ml-commons by opensearch-project.

the class MLCommonsRestTestCase method predict.

public void predict(RestClient client, FunctionName functionName, String modelId, String indexName, MLAlgoParams params, SearchSourceBuilder searchSourceBuilder, Consumer<Map<String, Object>> function) throws IOException {
    MLInputDataset inputData = SearchQueryInputDataset.builder().indices(ImmutableList.of(indexName)).searchSourceBuilder(searchSourceBuilder).build();
    MLInput kmeansInput = MLInput.builder().algorithm(functionName).parameters(params).inputDataset(inputData).build();
    String endpoint = "/_plugins/_ml/_predict/" + functionName.name().toLowerCase(Locale.ROOT) + "/" + modelId;
    Response response = TestHelper.makeRequest(client, "POST", endpoint, ImmutableMap.of(), TestHelper.toHttpEntity(kmeansInput), null);
    verifyResponse(function, response);
}