Examples with QuantileDiscretizerModel com.alibaba.alink.pipeline.feature.QuantileDiscretizerModel used on opensource projects

Example 1 with QuantileDiscretizerModel

use of com.alibaba.alink.pipeline.feature.QuantileDiscretizerModel in project Alink by alibaba.

the class PipelineSaveAndLoadTest method test2.

@Test
public void test2() throws Exception {
    String model_filename = "/tmp/model2.csv";
    CsvSourceBatchOp source = new CsvSourceBatchOp().setSchemaStr("sepal_length double, sepal_width double, petal_length double, petal_width double, category string").setFilePath("https://alink-test-data.oss-cn-hangzhou.aliyuncs.com/iris.csv");
    QuantileDiscretizerTrainBatchOp train = new QuantileDiscretizerTrainBatchOp().setNumBuckets(2).setSelectedCols("petal_length").linkFrom(source);
    train.link(new AkSinkBatchOp().setFilePath(model_filename).setOverwriteSink(true));
    BatchOperator.execute();
    // # save pipeline model data to file
    String pipelineModelFilename = "/tmp/model23424.csv";
    QuantileDiscretizer stage1 = new QuantileDiscretizer().setNumBuckets(2).setSelectedCols("sepal_length");
    Binarizer stage2 = new Binarizer().setSelectedCol("petal_width").setThreshold(1.);
    AkSourceBatchOp modelData = new AkSourceBatchOp().setFilePath(model_filename);
    QuantileDiscretizerModel stage3 = new QuantileDiscretizerModel().setSelectedCols("petal_length").setModelData(modelData);
    PipelineModel prevPipelineModel = new Pipeline(stage1, stage2, stage3).fit(source);
    prevPipelineModel.save(pipelineModelFilename, true);
    BatchOperator.execute();
}

Example 2 with QuantileDiscretizerModel

use of com.alibaba.alink.pipeline.feature.QuantileDiscretizerModel in project Alink by alibaba.

the class PipelineSaveAndLoadTest method testPipelineModelLoadSaveNested.

@Test
public void testPipelineModelLoadSaveNested() throws Exception {
    CsvSourceBatchOp source = new CsvSourceBatchOp().setSchemaStr("sepal_length double, sepal_width double, petal_length double, petal_width double, category string").setFilePath("https://alink-test-data.oss-cn-hangzhou.aliyuncs.com/iris.csv");
    String pipeline_model_filename = "/tmp/model123.csv";
    QuantileDiscretizerModel model1 = new QuantileDiscretizer().setNumBuckets(2).setSelectedCols("sepal_length").fit(source);
    Binarizer model2 = new Binarizer().setSelectedCol("petal_width").setThreshold(1.);
    PipelineModel pipeline_model = new PipelineModel(model1, model2);
    pipeline_model.save(pipeline_model_filename, true);
    BatchOperator.execute();
    pipeline_model = PipelineModel.load(pipeline_model_filename);
    BatchOperator<?> res = pipeline_model.transform(source);
    res.print();
}

Example 3 with QuantileDiscretizerModel

use of com.alibaba.alink.pipeline.feature.QuantileDiscretizerModel in project Alink by alibaba.

the class PipelineSaveAndLoadTest method test3.

@Test
public void test3() throws Exception {
    // # save model data to file (ModelBase)
    String modelFilename = "/tmp/model12341.csv";
    CsvSourceBatchOp source = new CsvSourceBatchOp().setSchemaStr("sepal_length double, sepal_width double, petal_length double, petal_width double, category string").setFilePath("https://alink-test-data.oss-cn-hangzhou.aliyuncs.com/iris.csv");
    new QuantileDiscretizer().setNumBuckets(2).setSelectedCols("petal_length").fit(source).getModelData().link(new CsvSinkBatchOp().setFilePath(modelFilename).setOverwriteSink(true));
    BatchOperator.execute();
    // # save pipeline model data to file
    QuantileDiscretizerModel model1 = new QuantileDiscretizer().setNumBuckets(2).setSelectedCols("sepal_length").fit(source);
    Binarizer model2 = new Binarizer().setSelectedCol("petal_width").setThreshold(1.);
    CsvSourceBatchOp modelData = new CsvSourceBatchOp().setFilePath(modelFilename).setSchemaStr("model_id BIGINT, model_info STRING");
    QuantileDiscretizerModel model3 = new QuantileDiscretizerModel().setSelectedCols("petal_length").setModelData(modelData);
    CsvSourceStreamOp streamSource = new CsvSourceStreamOp().setSchemaStr("sepal_length double, sepal_width double, petal_length double, petal_width double, category string").setFilePath("https://alink-test-data.oss-cn-hangzhou.aliyuncs.com/iris.csv");
    PipelineModel pipelineModel = new PipelineModel(model1, model2, model3);
    pipelineModel = PipelineModel.collectLoad(pipelineModel.save());
    pipelineModel.transform(streamSource).print();
    StreamOperator.execute();
}

Example 4 with QuantileDiscretizerModel

use of com.alibaba.alink.pipeline.feature.QuantileDiscretizerModel in project Alink by alibaba.

the class PipelineTest method test.

@Test
public void test() throws Exception {
    CsvSourceBatchOp source = new CsvSourceBatchOp().setSchemaStr("sepal_length double, sepal_width double, petal_length double, petal_width double, category string").setFilePath("https://alink-test-data.oss-cn-hangzhou.aliyuncs.com/iris.csv");
    String pipeline_model_filename = "/tmp/model123123123123.csv";
    QuantileDiscretizerModel model1 = new QuantileDiscretizer().setNumBuckets(2).setSelectedCols("sepal_length").fit(source);
    Binarizer model2 = new Binarizer().setSelectedCol("petal_width").setThreshold(1.);
    PipelineModel pipeline_model = new PipelineModel(model1, model2);
    pipeline_model.save(pipeline_model_filename, true);
    BatchOperator.execute();
    pipeline_model = PipelineModel.load(pipeline_model_filename);
    BatchOperator<?> res = pipeline_model.transform(source);
    res.print();
}