函數內的分組操作和跨多列的選項

Question

我想跨列對我使用的數據幀執行一些類似的操作。 請看鏈接獲取數據庫【非大文件】，由dput()生成

我想通過group_by(year,country_name_iso3)分組的cols= c("debt_GDP", "Top10")列執行以下三個操作：

1. inter = na.interpolation(cols, option = "spline")
2. sm=fitted(smooth.spline(cols_intep))
3. rollmean=rollmean(cols_intep,10, fill = NA)

這是一個包含單列的示例：

# interpolate missing values
df_us   <-subset(pkt, country_name_iso3=="USA")
df_us   <-droplevels(df_us)
df_us$debt_intep <-na.interpolation(df_us$debt_GDP, option = "spline")
df_us$top10_intep <-na.interpolation(df_us$Top10, option = "spline")

# smooth series with moving average 
df_us$debt_sm <- fitted(smooth.spline(df_us$debt_intep))
df_us$top10_sm <- fitted(smooth.spline(df_us$top10_intep))

# rolling mean 
df_us$debt_sm_rollmean<-rollmean(df_us$debt_sm,10, fill = NA)
df_us$top10_sm_rollmean<-rollmean(df_us$top10_sm,10, fill = NA)

對於按c(year,country_name_iso3)分組的每個c("debt_GDP", "Top10")列c("debt_GDP", "Top10")我希望完全相同

實現這一目標的最有效代碼是什么？

Answer 1

library(tidyverse)
library(imputeTS)
library(zoo)

使用mutate_at 3 次：

read_delim('dput.df.txt', delim = ' ') %>% 
  group_by(country_name_iso3) %>% 
  mutate_at(.vars = c('debt_GDP', 'Top10'), 
            .funs = list(inter = ~na_interpolation(., option = "spline"))) %>% 
  mutate_at(.vars = c('debt_GDP_inter', 'Top10_inter'), 
            .funs = list(sm = ~fitted(smooth.spline(.)))) %>% 
  mutate_at(.vars = c('debt_GDP_inter_sm', 'Top10_inter_sm'), 
            .funs = list(rollmean = ~rollmean(., 10, fill = NA)))

或者組合成一個函數：

func <- function(x) {
  inter = na_interpolation(x, option = 'spline')
  sm = fitted(smooth.spline(inter))
  rollmean = rollmean(sm, 10, fill = NA)
}


read_delim('dput.df.txt', delim = ' ') %>% 
  group_by(country_name_iso3) %>% 
  mutate_at(.vars = c('debt_GDP', 'Top10'), .funs = func)

繪制輸出

read_delim('dput.df.txt', delim = ' ') %>% 
  group_by(country_name_iso3) %>% 
  mutate_at(.vars = c('debt_GDP', 'Top10'), .funs = func) %>% 
  ggplot() +
  geom_line(aes(year, debt_GDP, color=country))