weibozhao commented on a change in pull request #24:
URL: https://github.com/apache/flink-ml/pull/24#discussion_r765459948
##########
File path:
flink-ml-lib/src/test/java/org/apache/flink/ml/classification/knn/KnnTest.java
##########
@@ -0,0 +1,273 @@
+package org.apache.flink.ml.classification.knn;
+
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.common.restartstrategy.RestartStrategies;
+import org.apache.flink.api.java.tuple.Tuple2;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.ml.linalg.DenseMatrix;
+import org.apache.flink.ml.linalg.DenseVector;
+import org.apache.flink.ml.linalg.Vectors;
+import org.apache.flink.ml.util.ReadWriteUtils;
+import org.apache.flink.streaming.api.datastream.DataStream;
+import
org.apache.flink.streaming.api.environment.ExecutionCheckpointingOptions;
+import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
+import org.apache.flink.table.api.DataTypes;
+import org.apache.flink.table.api.Schema;
+import org.apache.flink.table.api.Table;
+import org.apache.flink.table.api.bridge.java.StreamTableEnvironment;
+import org.apache.flink.table.api.internal.TableImpl;
+import org.apache.flink.types.Row;
+
+import org.apache.commons.collections.IteratorUtils;
+import org.junit.Assert;
+import org.junit.Before;
+import org.junit.Test;
+
+import java.nio.file.Files;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Objects;
+
+import static org.junit.Assert.assertEquals;
+
+/** Tests Knn and KnnModel. */
+public class KnnTest {
+ private StreamExecutionEnvironment env;
+ private StreamTableEnvironment tEnv;
+ private Table trainData;
+ private static final String LABEL_COL = "test_label";
+ private static final String PRED_COL = "test_prediction";
+ private static final String VEC_COL = "test_features";
+ private static final List<Row> trainArray =
+ new ArrayList<>(
+ Arrays.asList(
+ Row.of(1, Vectors.dense(2.0, 3.0)),
+ Row.of(1, Vectors.dense(2.1, 3.1)),
+ Row.of(2, Vectors.dense(200.1, 300.1)),
+ Row.of(2, Vectors.dense(200.2, 300.2)),
+ Row.of(2, Vectors.dense(200.3, 300.3)),
+ Row.of(2, Vectors.dense(200.4, 300.4)),
+ Row.of(2, Vectors.dense(200.4, 300.4)),
+ Row.of(2, Vectors.dense(200.6, 300.6)),
+ Row.of(1, Vectors.dense(2.1, 3.1)),
+ Row.of(1, Vectors.dense(2.1, 3.1)),
+ Row.of(1, Vectors.dense(2.1, 3.1)),
+ Row.of(1, Vectors.dense(2.1, 3.1)),
+ Row.of(1, Vectors.dense(2.3, 3.2)),
+ Row.of(1, Vectors.dense(2.3, 3.2)),
+ Row.of(3, Vectors.dense(2.8, 3.2)),
+ Row.of(4, Vectors.dense(300., 3.2)),
+ Row.of(1, Vectors.dense(2.2, 3.2)),
+ Row.of(5, Vectors.dense(2.4, 3.2)),
+ Row.of(5, Vectors.dense(2.5, 3.2)),
+ Row.of(5, Vectors.dense(2.5, 3.2)),
+ Row.of(1, Vectors.dense(2.1, 3.1))));
+
+ private static final List<Row> testArray =
+ new ArrayList<>(
+ Arrays.asList(
+ Row.of(5, Vectors.dense(4.0, 4.1)), Row.of(2,
Vectors.dense(300, 42))));
+ private Table testData;
+
+ @Before
+ public void before() {
+ Configuration config = new Configuration();
+
config.set(ExecutionCheckpointingOptions.ENABLE_CHECKPOINTS_AFTER_TASKS_FINISH,
true);
+ env = StreamExecutionEnvironment.getExecutionEnvironment(config);
+ env.setParallelism(4);
+ env.enableCheckpointing(100);
+ env.setRestartStrategy(RestartStrategies.noRestart());
+ tEnv = StreamTableEnvironment.create(env);
+
+ Schema schema =
+ Schema.newBuilder()
+ .column("f0", DataTypes.INT())
+ .column("f1", DataTypes.of(DenseVector.class))
+ .build();
+
+ DataStream<Row> dataStream = env.fromCollection(trainArray);
+ trainData = tEnv.fromDataStream(dataStream, schema).as(LABEL_COL + ","
+ VEC_COL);
+
+ DataStream<Row> predDataStream = env.fromCollection(testArray);
+ testData = tEnv.fromDataStream(predDataStream, schema).as(LABEL_COL +
"," + VEC_COL);
+ }
+
+ // Executes the graph and returns a list which has true label and predict
label.
+ private static List<Tuple2<String, String>> executeAndCollect(Table
output) throws Exception {
+ StreamTableEnvironment tEnv =
+ (StreamTableEnvironment) ((TableImpl)
output).getTableEnvironment();
+
+ DataStream<Tuple2<Integer, Integer>> stream =
+ tEnv.toDataStream(output)
+ .map(
+ new MapFunction<Row, Tuple2<Integer,
Integer>>() {
+ @Override
+ public Tuple2<Integer, Integer> map(Row
row) {
+ return Tuple2.of(
+ (Integer)
row.getField(LABEL_COL),
+ (Integer)
row.getField(PRED_COL));
+ }
+ });
+ return IteratorUtils.toList(stream.executeAndCollect());
+ }
+
+ private static void verifyClusteringResult(List<Tuple2<String, String>>
result) {
+ for (Tuple2<String, String> t2 : result) {
+ Assert.assertEquals(t2.f0, t2.f1);
+ }
+ }
+
+ /** Tests Param. */
+ @Test
+ public void testParam() {
+ Knn knn = new Knn();
+ assertEquals("features", knn.getFeaturesCol());
+ assertEquals("label", knn.getLabelCol());
+ assertEquals(10L, knn.getK().longValue());
+ assertEquals("prediction", knn.getPredictionCol());
+
+
knn.setLabelCol(LABEL_COL).setFeaturesCol(VEC_COL).setK(4).setPredictionCol(PRED_COL);
+
+ assertEquals(VEC_COL, knn.getFeaturesCol());
+ assertEquals(LABEL_COL, knn.getLabelCol());
+ assertEquals(4L, knn.getK().longValue());
+ assertEquals(PRED_COL, knn.getPredictionCol());
+ }
+
+ @Test
+ public void testFeaturePredictionParam() throws Exception {
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+ KnnModel model = knn.fit(trainData);
+ Table output = model.transform(testData)[0];
+
+ assertEquals(
+ Arrays.asList(LABEL_COL, VEC_COL, PRED_COL),
+ output.getResolvedSchema().getColumnNames());
+
+ List<Tuple2<String, String>> result = executeAndCollect(output);
+ verifyClusteringResult(result);
+ }
+
+ @Test
+ public void testFewerDistinctPointsThanCluster() throws Exception {
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+ KnnModel model = knn.fit(testData);
+ Table output = model.transform(testData)[0];
+
+ assertEquals(
+ Arrays.asList(LABEL_COL, VEC_COL, PRED_COL),
+ output.getResolvedSchema().getColumnNames());
+ executeAndCollect(output);
+ }
+
+ @Test
+ public void testFitAndPredict() throws Exception {
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+ KnnModel knnModel = knn.fit(trainData);
+ Table output = knnModel.transform(testData)[0];
+ List<Tuple2<String, String>> result = executeAndCollect(output);
+ verifyClusteringResult(result);
+ }
+
+ @Test
+ public void testSaveLoadAndPredict() throws Exception {
+ String path = Files.createTempDirectory("").toString();
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+ knn.save(path);
+
+ Knn loadKnn = Knn.load(path);
+ KnnModel knnModel = loadKnn.fit(trainData);
+ Table output = knnModel.transform(testData)[0];
+
+ List<Tuple2<String, String>> result = executeAndCollect(output);
+ verifyClusteringResult(result);
+ }
+
+ @Test
+ public void testModelSaveLoadAndPredict() throws Exception {
+ String path = Files.createTempDirectory("").toString();
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+ KnnModel knnModel = knn.fit(trainData);
+ knnModel.save(path);
+ env.execute();
+
+ KnnModel newModel = KnnModel.load(env, path);
+ Table output = newModel.transform(testData)[0];
+ List<Tuple2<String, String>> result = executeAndCollect(output);
+ verifyClusteringResult(result);
+ }
+
+ @Test
+ public void testGetModelData() throws Exception {
+ Knn knn =
+ new Knn()
+ .setLabelCol(LABEL_COL)
+ .setFeaturesCol(VEC_COL)
+ .setK(4)
+ .setPredictionCol(PRED_COL);
+
+ KnnModel knnModel = knn.fit(trainData);
+ Table modelData = knnModel.getModelData()[0];
+
+ DataStream<Row> output = tEnv.toDataStream(modelData);
+
+ assertEquals(
+ Arrays.asList("VECTORS", "NORM", "LABEL"),
+ modelData.getResolvedSchema().getColumnNames());
+
+ List<Row> modelRows = IteratorUtils.toList(output.executeAndCollect());
+ for (Row modelRow : modelRows) {
+ DenseMatrix vectors = (DenseMatrix) modelRow.getField(0);
+ DenseVector label = (DenseVector)
Objects.requireNonNull(modelRow.getField(1));
+
+ assertEquals(2, Objects.requireNonNull(vectors).numRows);
Review comment:
OK
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
