1

我正在努力为 Pig 的分组问题找到一些解决方案。目前我有一个看起来像的数据集；

Group | Height | Weight
  A   | 96.5   | 110.2
  B   | 88.2   | 122.5
  A   | 94.1   | 100.8
  B   | 84.1   | 115.6

我正在使用 DataFu 库中的 StreamingQuantile 方法来计算高度变量的分位数（第 25、第 50 ... 蚀刻）。目前它有效，但我还需要计算每个组的 AVG 权重 + 它们的分位数；所以它看起来像这样；

A | Quantile1 | 88.5 (height)  | 134.4 (avg weight)
A | Quantile2 | 125.3 (height) | 156.2 (avg weight)
etc.....
B | Quantile4 | 144.3 (height) | 134.2 (avg weight)

作为参考，这里是用于计算分位数的简单 Pig；

REGISTER /usr/lib/datafu-1.2.0.jar;
define Quantile datafu.pig.stats.StreamingQuantile('0.0','0.25','0.5','0.75','1.0');
A = load 'mydata';
Group_A = GROUP A BY $0;
Quant = FOREACH GROUP_A GENERATE group,Quantile(A.$1);

无论如何我也可以计算每个分位数和组的平均值 2 美元吗？

4

1 回答 1

0

我为这些列命名以避免混淆。

REGISTER /usr/lib/datafu-1.2.0.jar;
define Quantile datafu.pig.stats.StreamingQuantile('0.0','0.25','0.5','0.75','1.0');
A = load 'mydata' using PigStorage('|') as (grouping:chararray, height:double, weight:double);
Group_A = GROUP A BY grouping;
Quant = FOREACH GROUP_A GENERATE group, Quantile(A.height) as q;
joined = join Quant by group, A by grouping;
grouped = group joined by (grouping, q)
average = foreach grouped generate group, AVG(joined.weight);
于 2014-09-11T17:53:16.163 回答