• ALINK(三十八):模型评估(三)多分类评估 (EvalMultiClassBatchOp)


    Java 类名:com.alibaba.alink.operator.batch.evaluation.EvalMultiClassBatchOp

    Python 类名:EvalMultiClassBatchOp

    功能介绍

    多分类评估是对多分类算法的预测结果进行效果评估。

    支持Roc曲线,LiftChart曲线,K-S曲线,Recall-Precision曲线绘制。

    流式的实验支持累计统计和窗口统计,除却上述四条曲线外,还给出Auc/Kappa/Accuracy/Logloss随时间的变化曲线。

    给出整体的评估指标包括:AUC、K-S、PRC, 不同阈值下的Precision、Recall、F-Measure、Sensitivity、Accuracy、Specificity和Kappa。

    混淆矩阵

     

    F-Measure

    F1=dfrac{2TP}{2TP+FP+FN}=dfrac{2cdot Precision cdot Recall}{Precision+Recall}

     

    参数说明

    名称

    中文名称

    描述

    类型

    是否必须?

    默认值

    labelCol

    标签列名

    输入表中的标签列名

    String

     

    predictionCol

    预测结果列名

    预测结果列名

    String

       

    predictionDetailCol

    预测详细信息列名

    预测详细信息列名

    String

       

    代码示例

    Python 代码

    from pyalink.alink import *
    import pandas as pd
    useLocalEnv(1)
    df = pd.DataFrame([
        ["prefix1", "{"prefix1": 0.9, "prefix0": 0.1}"],
        ["prefix1", "{"prefix1": 0.8, "prefix0": 0.2}"],
        ["prefix1", "{"prefix1": 0.7, "prefix0": 0.3}"],
        ["prefix0", "{"prefix1": 0.75, "prefix0": 0.25}"],
        ["prefix0", "{"prefix1": 0.6, "prefix0": 0.4}"]])
    inOp = BatchOperator.fromDataframe(df, schemaStr='label string, detailInput string')
    metrics = EvalMultiClassBatchOp().setLabelCol("label").setPredictionDetailCol("detailInput").linkFrom(inOp).collectMetrics()
    print("Prefix0 accuracy:", metrics.getAccuracy("prefix0"))
    print("Prefix1 recall:", metrics.getRecall("prefix1"))
    print("Macro Precision:", metrics.getMacroPrecision())
    print("Micro Recall:", metrics.getMicroRecall())
    print("Weighted Sensitivity:", metrics.getWeightedSensitivity())

    Java 代码

    import org.apache.flink.types.Row;
    import com.alibaba.alink.operator.batch.BatchOperator;
    import com.alibaba.alink.operator.batch.evaluation.EvalMultiClassBatchOp;
    import com.alibaba.alink.operator.batch.source.MemSourceBatchOp;
    import com.alibaba.alink.operator.common.evaluation.MultiClassMetrics;
    import org.junit.Test;
    import java.util.Arrays;
    import java.util.List;
    public class EvalMultiClassBatchOpTest {
      @Test
      public void testEvalMultiClassBatchOp() throws Exception {
        List <Row> df = Arrays.asList(
          Row.of("prefix1", "{"prefix1": 0.9, "prefix0": 0.1}"),
          Row.of("prefix1", "{"prefix1": 0.8, "prefix0": 0.2}"),
          Row.of("prefix1", "{"prefix1": 0.7, "prefix0": 0.3}"),
          Row.of("prefix0", "{"prefix1": 0.75, "prefix0": 0.25}")
        );
        BatchOperator <?> inOp = new MemSourceBatchOp(df, "label string, detailInput string");
        MultiClassMetrics metrics = new EvalMultiClassBatchOp().setLabelCol("label").setPredictionDetailCol(
          "detailInput").linkFrom(inOp).collectMetrics();
        System.out.println("Prefix0 accuracy:" + metrics.getAccuracy("prefix0"));
        System.out.println("Prefix1 recall:" + metrics.getRecall("prefix1"));
        System.out.println("Macro Precision:" + metrics.getMacroPrecision());
        System.out.println("Micro Recall:" + metrics.getMicroRecall());
        System.out.println("Weighted Sensitivity:" + metrics.getWeightedSensitivity());
      }
    }

    运行结果

    Prefix0 accuracy: 0.6
    Prefix1 recall: 1.0
    Macro Precision: 0.8
    Micro Recall: 0.6
    Weighted Sensitivity: 0.6
  • 相关阅读:
    Js五子棋游戏
    换肤
    jqzoom实战
    iphone开发笔记
    腾讯召实习生笔试题
    ajax( 汇率转化)
    20140805&nbsp;17:24
    加载上一级目录下的图片文件并得到加载图片的大小
    精确碰撞检测 bitmapData.hitTest
    buttonMode 与mouseChildren【替换鼠标之后不响应鼠标事件】
  • 原文地址:https://www.cnblogs.com/qiu-hua/p/14902345.html
Copyright © 2020-2023  润新知