Java 类名：com.alibaba.alink.operator.batch.feature.EqualWidthDiscretizerTrainBatchOp
Python 类名：EqualWidthDiscretizerTrainBatchOp

功能介绍

等宽离散可以计算选定数值列的分位点，每个区间都有相同的组距，也就是数据范围/组数，通过训练可以得到一系列分为点，
然后使用这些分位点进行预测。
其中可以所有列使用同一个分组数量，也可以每一列对应一个分组数量。预测结果可以是特征值或一系列0/1离散特征。

参数说明

名称 中文名称 描述 类型 是否必须？ 取值范围 默认值
selectedCols 选择的列名 计算列对应的列名列表 String[] ✓ 所选列类型为 [BIGDECIMAL, BIGINTEGER, BYTE, DOUBLE, FLOAT, INTEGER, LONG, SHORT]
leftOpen 是否左开右闭 左开右闭为true，左闭右开为false Boolean true
numBuckets quantile个数 quantile个数，对所有列有效。 Integer 2
numBucketsArray quantile个数 quantile个数，每一列对应数组中一个元素。 Integer[] null

代码示例

Python 代码

  1. from pyalink.alink import *
  3. import pandas as pd
  5. useLocalEnv(1)
  7. df = pd.DataFrame([
  8.     ["a", 1, 1.1],     
  9.     ["b", -2, 0.9],    
  10.     ["c", 100, -0.01], 
  11.     ["d", -99, 100.9], 
  12.     ["a", 1, 1.1],     
  13.     ["b", -2, 0.9],    
  14.     ["c", 100, -0.01], 
  15.     ["d", -99, 100.9] 
  16. ])
  18. batchSource =  BatchOperator.fromDataframe(df,schemaStr="f_string string, f_long long, f_double double")
  20. trainOp = EqualWidthDiscretizerTrainBatchOp(). \
  21. setSelectedCols(['f_long', 'f_double']). \
  22. setNumBuckets(5). \
  23. linkFrom(batchSource)
  25. EqualWidthDiscretizerPredictBatchOp(). \
  26. setSelectedCols(['f_long', 'f_double']). \
  27. linkFrom(trainOp,batchSource). \
  28. print()
  30. trainOp = EqualWidthDiscretizerTrainBatchOp().setSelectedCols(['f_long', 'f_double']). \
  31. setNumBucketsArray([5,3]). \
  32. linkFrom(batchSource)
  34. EqualWidthDiscretizerPredictBatchOp(). \
  35. setSelectedCols(['f_long', 'f_double']). \
  36. linkFrom(trainOp,batchSource). \
  37. print()
  39. EqualWidthDiscretizerPredictBatchOp(). \
  40. setEncode("ASSEMBLED_VECTOR"). \
  41. setSelectedCols(['f_long', 'f_double']). \
  42. setOutputCols(["assVec"]). \
  43. linkFrom(trainOp,batchSource).print()

Java 代码

  1. import org.apache.flink.types.Row;
  3. import com.alibaba.alink.operator.batch.BatchOperator;
  4. import com.alibaba.alink.operator.batch.feature.EqualWidthDiscretizerPredictBatchOp;
  5. import com.alibaba.alink.operator.batch.feature.EqualWidthDiscretizerTrainBatchOp;
  6. import com.alibaba.alink.operator.batch.source.MemSourceBatchOp;
  7. import com.alibaba.alink.operator.stream.StreamOperator;
  8. import com.alibaba.alink.operator.stream.feature.EqualWidthDiscretizerPredictStreamOp;
  9. import com.alibaba.alink.operator.stream.source.MemSourceStreamOp;
  10. import com.alibaba.alink.params.feature.HasEncodeWithoutWoe.Encode;
  11. import org.junit.Test;
  13. import java.util.Arrays;
  14. import java.util.List;
  16. public class EqualWidthDiscretizerTrainBatchOpTest {
  17.     @Test
  18.     public void testEqualWidthDiscretizerTrainBatchOp2() throws Exception {
  19.         List <Row> df = Arrays.asList(
  20.             Row.of("a", 1, 1.1),
  21.             Row.of("b", -2, 0.9),
  22.             Row.of("c", 100, -0.01),
  23.             Row.of("d", -99, 100.9),
  24.             Row.of("a", 1, 1.1),
  25.             Row.of("b", -2, 0.9),
  26.             Row.of("c", 100, -0.01),
  27.             Row.of("d", -99, 100.9)
  28.         );
  29.         BatchOperator <?> batchSource = new MemSourceBatchOp(df, "f_string string, f_long int, f_double double");
  31.         BatchOperator <?> trainOp = new EqualWidthDiscretizerTrainBatchOp().setSelectedCols("f_long", "f_double")
  32.             .setNumBuckets(5).linkFrom(batchSource);
  34.         new EqualWidthDiscretizerPredictBatchOp().setSelectedCols("f_long","f_double")
  35.             .linkFrom(trainOp, batchSource).print();
  37.         BatchOperator trainOp2 = new EqualWidthDiscretizerTrainBatchOp().setSelectedCols("f_long", "f_double")
  38.             .setNumBucketsArray(5,3).linkFrom(batchSource);
  40.         new EqualWidthDiscretizerPredictBatchOp().setSelectedCols("f_long","f_double")
  41.             .linkFrom(trainOp2,batchSource).print();
  43.         new EqualWidthDiscretizerPredictBatchOp().setSelectedCols("f_long","f_double")
  44.             .setEncode(Encode.ASSEMBLED_VECTOR)
  45.             .setOutputCols("assVec")
  46.             .linkFrom(trainOp2,batchSource).print();
  47.     }
  48. }

运行结果

| f_string | f_long | f_double | | —- | —- | —- |

| a | 2 | 0 |

| b | 2 | 0 |

| c | 4 | 0 |

| d | 0 | 4 |

| a | 2 | 0 |

| b | 2 | 0 |

| c | 4 | 0 |

| d | 0 | 4 |

| f_string | f_long | f_double | | —- | —- | —- |

| a | 2 | 0 |

| b | 2 | 0 |

| c | 4 | 0 |

| d | 0 | 2 |

| a | 2 | 0 |

| b | 2 | 0 |

| c | 4 | 0 |

| d | 0 | 2 |

| f_string | f_long | f_double | assVec | | —- | —- | —- | —- |

| a | 1 | 1.1000 | $8$2:1.0 5:1.0 |

| b | -2 | 0.9000 | $8$2:1.0 5:1.0 |

| c | 100 | -0.0100 | $8$5:1.0 |

| d | -99 | 100.9000 | $8$0:1.0 |

| a | 1 | 1.1000 | $8$2:1.0 5:1.0 |

| b | -2 | 0.9000 | $8$2:1.0 5:1.0 |

| c | 100 | -0.0100 | $8$5:1.0 |

| d | -99 | 100.9000 | $8$0:1.0 |