本章包括下面各节：
12.1 多分类模型评估方法
12.1.1 综合指标
12.1.2 关于每个标签值的二分类指标
12.1.3 Micro、Macro、Weighted计算的指标
12.2 数据探索
12.3 使用朴素贝叶斯进行多分类
12.4 二分类器组合
12.5 Softmax算法
12.6 多层感知器分类器
详细内容请阅读纸质书《Alink权威指南：基于Flink的机器学习实例入门（Java）》，这里为本章对应的示例代码。
package com.alibaba.alink;
import com.alibaba.alink.operator.batch.BatchOperator;
import com.alibaba.alink.operator.batch.classification.NaiveBayesPredictBatchOp;
import com.alibaba.alink.operator.batch.classification.NaiveBayesTrainBatchOp;
import com.alibaba.alink.operator.batch.evaluation.EvalMultiClassBatchOp;
import com.alibaba.alink.operator.batch.source.AkSourceBatchOp;
import com.alibaba.alink.operator.batch.source.CsvSourceBatchOp;
import com.alibaba.alink.operator.batch.statistics.CorrelationBatchOp;
import com.alibaba.alink.pipeline.classification.GbdtClassifier;
import com.alibaba.alink.pipeline.classification.LinearSvm;
import com.alibaba.alink.pipeline.classification.LogisticRegression;
import com.alibaba.alink.pipeline.classification.MultilayerPerceptronClassifier;
import com.alibaba.alink.pipeline.classification.OneVsRest;
import com.alibaba.alink.pipeline.classification.Softmax;
import java.io.File;
public class Chap12 {
static final String DATA_DIR = Utils.ROOT_DIR + "iris" + File.separator;
static final String ORIGIN_FILE = "iris.data";
static final String TRAIN_FILE = "train.ak";
static final String TEST_FILE = "test.ak";
static final String SCHEMA_STRING
= "sepal_length double, sepal_width double, petal_length double, petal_width double, category string";
static final String[] FEATURE_COL_NAMES
= new String[] {"sepal_length", "sepal_width", "petal_length", "petal_width"};
static final String LABEL_COL_NAME = "category";
static final String PREDICTION_COL_NAME = "pred";
static final String PRED_DETAIL_COL_NAME = "pred_info";
public static void main(String[] args) throws Exception {
BatchOperator.setParallelism(1);
c_2();
c_3();
c_4();
c_5();
c_6();
}
static void c_2() throws Exception {
CsvSourceBatchOp source =
new CsvSourceBatchOp()
.setFilePath(DATA_DIR + ORIGIN_FILE)
.setSchemaStr(SCHEMA_STRING);
source
.lazyPrint(5, "origin file")
.lazyPrintStatistics("stat of origin file")
.link(
new CorrelationBatchOp()
.setSelectedCols(FEATURE_COL_NAMES)
.lazyPrintCorrelation()
);
source.groupBy(LABEL_COL_NAME, LABEL_COL_NAME + ", COUNT(*) AS cnt").lazyPrint(-1);
BatchOperator.execute();
Utils.splitTrainTestIfNotExist(source, DATA_DIR + TRAIN_FILE, DATA_DIR + TEST_FILE, 0.9);
}
static void c_3() throws Exception {
AkSourceBatchOp train_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TRAIN_FILE);
AkSourceBatchOp test_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TEST_FILE);
NaiveBayesTrainBatchOp trainer =
new NaiveBayesTrainBatchOp()
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME);
NaiveBayesPredictBatchOp predictor =
new NaiveBayesPredictBatchOp()
.setPredictionCol(PREDICTION_COL_NAME)
.setPredictionDetailCol(PRED_DETAIL_COL_NAME);
train_data.link(trainer);
predictor.linkFrom(trainer, test_data);
trainer.lazyPrintModelInfo();
predictor.lazyPrint(1, "< Prediction >");
predictor
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.setPredictionDetailCol(PRED_DETAIL_COL_NAME)
.lazyPrintMetrics("NaiveBayes")
);
BatchOperator.execute();
}
static void c_4() throws Exception {
AkSourceBatchOp train_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TRAIN_FILE);
AkSourceBatchOp test_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TEST_FILE);
new OneVsRest()
.setClassifier(
new LogisticRegression()
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
)
.setNumClass(3)
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("OneVsRest_LogisticRegression")
);
new OneVsRest()
.setClassifier(
new GbdtClassifier()
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
)
.setNumClass(3)
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("OneVsRest_GBDT")
);
new OneVsRest()
.setClassifier(
new LinearSvm()
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
)
.setNumClass(3)
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("OneVsRest_LinearSvm")
);
BatchOperator.execute();
}
static void c_5() throws Exception {
AkSourceBatchOp train_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TRAIN_FILE);
AkSourceBatchOp test_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TEST_FILE);
new Softmax()
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.enableLazyPrintTrainInfo()
.enableLazyPrintModelInfo()
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("Softmax")
);
BatchOperator.execute();
}
static void c_6() throws Exception {
AkSourceBatchOp train_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TRAIN_FILE);
AkSourceBatchOp test_data = new AkSourceBatchOp().setFilePath(DATA_DIR + TEST_FILE);
new MultilayerPerceptronClassifier()
.setLayers(new int[] {4, 12, 3})
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("MultilayerPerceptronClassifier [4, 12, 3]")
);
new MultilayerPerceptronClassifier()
.setLayers(new int[] {4, 3})
.setFeatureCols(FEATURE_COL_NAMES)
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.fit(train_data)
.transform(test_data)
.link(
new EvalMultiClassBatchOp()
.setLabelCol(LABEL_COL_NAME)
.setPredictionCol(PREDICTION_COL_NAME)
.lazyPrintMetrics("MultilayerPerceptronClassifier [4, 3]")
);
BatchOperator.execute();
}
}