Условные средства на основе других столбцов в R с dplyr

Question

Допустим, у меня есть следующие данные:

structure(list(political_spectrum = c(5L, 15L, 12L, 30L, 100L, 
0L, 27L, 52L, 38L, 64L, 0L, 0L, 76L, 50L, 16L, 16L, 0L, 23L, 
0L, 25L, 68L, 50L, 4L, 0L, 50L), politics_today = c("Independent", 
"Strong Democrat", "Weak Democrat", "Weak Democrat", "Weak Republican", 
"Strong Democrat", "Weak Democrat", "Weak Democrat", "Independent", 
"Weak Democrat", "Strong Democrat", "Independent", "Weak Republican", 
"Weak Democrat", "Weak Democrat", "Strong Democrat", "Strong Democrat", 
"Strong Democrat", "Strong Democrat", "Strong Democrat", "Independent", 
"Independent", "Strong Democrat", "Strong Democrat", "Independent"
), stranger_things_universe_mc = c("The Demagorgon", "", "", 
"", "", "", "", "", "", "The Stranger Land", "The Demagorgon", 
"The Upside Down", "", "", "", "", "", "The Upside Down", "The Shadowland", 
"", "", "", "", "", "The Shadowland"), stranger_things_universe_answer = c("The Upside Down", 
"", "", "", "", "", "", "", "", "The Upside Down", "The Upside Down", 
"The Upside Down", "", "", "", "", "", "The Upside Down", "The Upside Down", 
"", "", "", "", "", "The Upside Down"), stranger_things_universe_confidence = c(32L, 
NA, NA, NA, NA, NA, NA, NA, NA, 67L, 94L, 89L, NA, NA, NA, NA, 
NA, 51L, 10L, NA, NA, NA, NA, NA, 0L), stranger_things_universe_importance = c("Don't care at all", 
"", "", "", "", "", "", "", "", "Care somewhat strongly", "Care a little", 
"Care somewhat strongly", "", "", "", "", "", "Care somewhat", 
"Don't care at all", "", "", "", "", "", "Don't care at all"), 
    tupac_mc = c("", "Biggie Smalls", "", "", "", "", "", "Biggie Smalls", 
    "Biggie Smalls", "", "", "Biggie Smalls", "", "", "", "", 
    "", "", "Biggie Smalls", "", "", "Ice Cube", "", "", ""), 
    tupac_answer = c("", "Biggie Smalls", "", "", "", "", "", 
    "Biggie Smalls", "Biggie Smalls", "", "", "Biggie Smalls", 
    "", "", "", "", "", "", "Biggie Smalls", "", "", "Biggie Smalls", 
    "", "", ""), tupac_confidence = c(NA, 70L, NA, NA, NA, NA, 
    NA, 71L, 76L, NA, NA, 100L, NA, NA, NA, NA, NA, NA, 100L, 
    NA, NA, 32L, NA, NA, NA), tupac_importance = c("", "Don't care at all", 
    "", "", "", "", "", "Care somewhat", "Don't care at all", 
    "", "", "Care strongly", "", "", "", "", "", "", "Care a little", 
    "", "", "Don't care at all", "", "", ""), uber_ceo_mc = c("John Zimmer", 
    "", "", "", "", "Travis Kalanick", "", "", "", "Travis Kalanick", 
    "", "", "", "", "", "", "", "John Zimmer", "Travis Kalanick", 
    "Travis Kalanick", "", "", "", "", ""), uber_ceo_answer = c("Travis Kalanick", 
    "", "", "", "", "Travis Kalanick", "", "", "", "Travis Kalanick", 
    "", "", "", "", "", "", "", "Travis Kalanick", "Travis Kalanick", 
    "Travis Kalanick", "", "", "", "", ""), uber_ceo_confidence = c(0L, 
    NA, NA, NA, NA, 94L, NA, NA, NA, 69L, NA, NA, NA, NA, NA, 
    NA, NA, 5L, 13L, 17L, NA, NA, NA, NA, NA), uber_ceo_importance = c("Don't care at all", 
    "", "", "", "", "Care strongly", "", "", "", "Care somewhat", 
    "", "", "", "", "", "", "", "Don't care at all", "Don't care at all", 
    "Care somewhat", "", "", "", "", ""), black_panther_mc = c("", 
    "T'Chaka", "", "", "", "", "", "", "", "", "", "", "", "", 
    "", "", "", "", "", "", "", "", "T'Chaka", "", ""), black_panther_answer = c("", 
    "T'Challa", "", "", "", "", "", "", "", "", "", "", "", "", 
    "", "", "", "", "", "", "", "", "T'Challa", "", ""), black_panther_confidence = c(NA, 
    63L, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
    NA, NA, NA, NA, NA, NA, NA, 34L, NA, NA), black_panther_importance = c("", 
    "Don't care at all", "", "", "", "", "", "", "", "", "", 
    "", "", "", "", "", "", "", "", "", "", "", "Care a little", 
    "", ""), the_office_mc = c("The Mindy Project", "", "", "", 
    "", "", "", "", "", "", "", "", "", "", "The Office", "", 
    "", "The Mindy Project", "", "", "", "", "The Office", "", 
    ""), the_office_answer = c("The Office", "", "", "", "", 
    "", "", "", "", "", "", "", "", "", "The Office", "", "", 
    "The Office", "", "", "", "", "The Office", "", ""), the_office_confidence = c(43L, 
    NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 2L, NA, 
    NA, 11L, NA, NA, NA, NA, 100L, NA, NA), the_office_importance = c("Don't care at all", 
    "", "", "", "", "", "", "", "", "", "", "", "", "", "Don't care at all", 
    "", "", "Care a little", "", "", "", "", "Care a little", 
    "", ""), arms_manufacturing_company_mc = c("J. Brockton & Sons", 
    "", "", "O.F. Mossberg & Sons", "", "", "", "", "", "", "", 
    "", "J. Brockton & Sons", "", "", "", "", "", "", "", "", 
    "", "", "", "J. Brockton & Sons"), arms_manufacturing_company_answer = c("J. Brockton & Sons", 
    "", "", "J. Brockton & Sons", "", "", "", "", "", "", "", 
    "", "J. Brockton & Sons", "", "", "", "", "", "", "", "", 
    "", "", "", "J. Brockton & Sons"), arms_manufacturing_company_confidence = c(91L, 
    NA, NA, 24L, NA, NA, NA, NA, NA, NA, NA, NA, 37L, NA, NA, 
    NA, NA, NA, NA, NA, NA, NA, NA, NA, 100L), arms_manufacturing_company_importance = c("Don't care at all", 
    "", "", "Don't care at all", "", "", "", "", "", "", "", 
    "", "Don't care at all", "", "", "", "", "", "", "", "", 
    "", "", "", "Don't care at all")), class = c("data.table", 
"data.frame"), row.names = c(NA, -25L))

Я пытаюсь сделать что-то вроде следующего:

test %>%
  gather(name, value, -c('political_spectrum', 'politics_today')) %>%
  filter(value != "") %>%
  mutate(question_id = sub("_[^_]+$", "", name)) %>%
  mutate(confidence = grepl("_confidence", name)) %>%
  group_by(politics_today, question_id) %>%
  summarize(mean_confidence = mean(value[confidence == "TRUE"]))

, в котором я получаю значения mean_confidence для каждой политической принадлежности, но только для определенных строк в столбце «значение». Чтобы запустить среднее значение только для столбцов «доверие», я пытаюсь выполнить фильтрацию через mean(value[confidence == "TRUE"]), но не уверен, что это можно сделать правильно.

Answer 1

Я думаю, вам нужно изменить код на

library(tidyverse)

test %>%
  gather(name, value, -c('political_spectrum', 'politics_today')) %>%
  filter(value != "") %>%
  mutate(question_id = sub("_[^_]+$", "", name),
         confidence = grepl("_confidence", name)) %>%
  group_by(politics_today, question_id) %>%
  summarize(mean_confidence = mean(as.numeric(value[confidence])))


#   politics_today  question_id                mean_confidence
#   <chr>           <chr>                                <dbl>
# 1 Independent     arms_manufacturing_company            95.5
# 2 Independent     stranger_things_universe              40.3
# 3 Independent     the_office                            43  
# 4 Independent     tupac                                 69.3
# 5 Independent     uber_ceo                               0  
# 6 Strong Democrat black_panther                         48.5
# 7 Strong Democrat stranger_things_universe              51.7
# 8 Strong Democrat the_office                            55.5
# 9 Strong Democrat tupac                                 85  
#10 Strong Democrat uber_ceo                              32.2
#11 Weak Democrat   arms_manufacturing_company            24  
#12 Weak Democrat   stranger_things_universe              67  
#13 Weak Democrat   the_office                             2  
#14 Weak Democrat   tupac                                 71  
#15 Weak Democrat   uber_ceo                              69  
#16 Weak Republican arms_manufacturing_company            37  

Поскольку ваш столбец value имеет как числовые, так и символьные значения, он преобразуется в символьный столбец, поэтому вам нужно изменить value где confidence == TRUE к числовому.