Rowsums以循环中的列名为条件 [英] Rowsums conditional on column name in a loop

问题描述

这是一则后续问题： Rowsums以列名称为条件

我的数据框称为 wiod ，如下所示： p>

My data frame is called wiod and looks like this:

VAR1 VAR2 AUS1 ... AUS56 BEL1 ... BEL56 NLD1 ... NLD56
A    D    23   ... 99    0    ... 444   123  ... 675
B    D    55   ... 6456  0    ... 557   567  ... 4345

我想计算变量 AUS，BEL，NLD 的行总和，然后删除旧的变量。像这样：

I'd like to calculate the row-sums for the variables AUS, BEL, NLD and then drop the old variables. Like this:

wiot <- wiot %>% 
  mutate(AUS = rowSums(.[grep("AUS", names(.))])) %>% 
  mutate(BEL = rowSums(.[grep("BEL", names(.))])) %>% 
  mutate(NLD = rowSums(.[grep("NLD", names(.))])) %>% 
  select(Var1, Var2, AUS, BEL, NLD)

当然，有大量的变量组，不只是这三个（43，要精确）。有没有任何方便的方法，而不使用43变异命令？

Of course, there is a large number of the variable groups, not just these three (43, to be precise). Is there any convenient way to do this without using 43 mutate commands?

推荐答案

它使得更容易从宽格式转换为长（收集），然后总结，如果需要转换回广泛（传播）格式：

It makes it easier to convert from wide format to long (gather), then summarise, and if needed convert back to wide (spread) format:

library(dplyr)
library(tidyr)

# dataframe from @989 http://stackoverflow.com/a/43519062
df1 %>% 
  gather(key = myKey, value = myValue, -c(VAR1, VAR2)) %>% 
  mutate(myGroup = gsub("\\d", "", myKey)) %>% 
  group_by(VAR1, VAR2, myGroup) %>% 
  summarise(mySum = sum(myValue)) %>% 
  spread(key = myGroup, value = mySum)

# Source: local data frame [2 x 5]
# Groups: VAR1, VAR2 [2]
# 
#     VAR1   VAR2   AUS   BEL   NLD
# * <fctr> <fctr> <int> <int> <int>
# 1      A      D   122   444   798
# 2      B      D  6511   557  4912

这篇关于Rowsums以循环中的列名为条件的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！