什么聚合函数可以与sparklyr中的sdf_pivot一起使用？

Question

尝试将sdf_pivot与sdf_pivot的开发版本sparklyr 。 唯一可行的聚合函数是count 。 如果我尝试sum或avg我得到一个异常，说明No matched method found for class org.apache.spark.sql.RelationalGroupedDataset.sum

这是一些重现的代码：

iris_tbl <- copy_to(sc, iris)
iris_tbl %>% sdf_pivot(Species ~ Sepal_Width) # this works
iris_tbl %>% sdf_pivot(Species ~ Sepal_Width, "sum") # this doesn't 

Answer 1

我相信这仍然没有记录，但是你得到这个错误的原因是你需要使用带有R列表或R函数的sdf_pivot 函数进行聚合 。

以下是一些例子：

使用R列表：

> iris_tbl %>% sdf_pivot(Species ~ Sepal_Width, list(Sepal_Width="sum")) %>% head()
# Source:   lazy query [?? x 24]
# Database: spark_connection
     Species `2.0` `2.2` `2.3` `2.4` `2.5` `2.6` `2.7` `2.8` `2.9` `3.0` `3.1` `3.2` `3.3`
       <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 versicolor     2   4.4   6.9   7.2    10   7.8  13.5  16.8  20.3    24   9.3   9.6   3.3
2  virginica   NaN   2.2   NaN   NaN    10   5.2  10.8  22.4   5.8    36  12.4  16.0   9.9
3     setosa   NaN   NaN   2.3   NaN   NaN   NaN   NaN   NaN   2.9    18  12.4  16.0   6.6
# ... with 10 more variables: `3.4` <dbl>, `3.5` <dbl>, `3.6` <dbl>, `3.7` <dbl>,
#   `3.8` <dbl>, `3.9` <dbl>, `4.0` <dbl>, `4.1` <dbl>, `4.2` <dbl>, `4.4` <dbl>

使用R功能：

> sum_sepal_width <- function(gdf) {
  expr <- invoke_static(
          sc,
          "org.apache.spark.sql.functions",
          "expr",
          "sum(Sepal_Width)"
      )

   gdf %>% invoke("agg", expr, list())
}

> iris_tbl %>% sdf_pivot(Species ~ Sepal_Width, fun.aggregate = fun.aggregate)
# Source:   table<sparklyr_tmp_4ee61c86311c> [?? x 24]
# Database: spark_connection
     Species `2.0` `2.2` `2.3` `2.4` `2.5` `2.6` `2.7` `2.8` `2.9` `3.0` `3.1` `3.2` `3.3`
       <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 versicolor     2   4.4   6.9   7.2    10   7.8  13.5  16.8  20.3    24   9.3   9.6   3.3
2  virginica   NaN   2.2   NaN   NaN    10   5.2  10.8  22.4   5.8    36  12.4  16.0   9.9
3     setosa   NaN   NaN   2.3   NaN   NaN   NaN   NaN   NaN   2.9    18  12.4  16.0   6.6
# ... with 10 more variables: `3.4` <dbl>, `3.5` <dbl>, `3.6` <dbl>, `3.7` <dbl>,
#   `3.8` <dbl>, `3.9` <dbl>, `4.0` <dbl>, `4.1` <dbl>, `4.2` <dbl>, `4.4` <dbl>

注意：在sparklyr-0-6-0- sdf_pivot之前， sdf_pivot不可用。