Delete rows based on multiple conditions with dply

问题:

I am trying to delete specific rows in my dataset based on values in multiple columns. A row should be deleted only when a condition in all 3 columns is met.

This is my code:

test_dff %>%
  filter(contbr_nm != c('GAITHER, BARBARA', 'PANIC, RADIVOJE', 'KHAN, RAMYA') & 
           contbr_city != c('APO AE', 'PORSGRUNN', 'NEW YORK') &
           contbr_zip != c('9309', '3924', '2586'))

This code should remove 12 rows in my table. Instead it removes a vast majority of them. I am suspecting, that it removes all the possible rows, whenever one of the conditions is met.

Is there a better solution, or do I have to use the approach, described here?

Do I need to specify each combination separately? Like so? This approach also deletes far too many rows, so it is also wrong.

test_dff %>%
  filter((contbr_nm != 'GAITHER, BARBARA' & contbr_city != 'APO AE' & contbr_zip != '9309') &
         (contbr_nm != 'PANIC, RADIVOJE' & contbr_city != 'PORSGRUNN' & contbr_zip != '3924') &
           (contbr_nm != 'KHAN, RAMYA' & contbr_city != 'NEW YORK' & contbr_zip != '2586') )

If I focus on deleting rows only based on one variable, this piece of code works:

test_dff %>%
  filter(contbr_zip != c('9309')) %>%
  filter(contbr_zip != c('3924')) %>%
  filter(contbr_zip != c('2586'))

Why does such an approach not work?

test_dff %>%
  filter(contbr_zip != c('9309','3924','2586')) 

Thanks a lot for your help.

回答1:

Here is a join-based approach - all items must be exact matches.

main <- read.csv(text = "
id,name,city,zip
1,mary,new york,10017
2,jonah,new york,10016
3,tamil,manhattan,10019
4,vijay,harlem,10028
")

excludes <- read.csv(text = "
name,city,zip
jonah,new york,10016
vijay,harlem,10028
")

library(dplyr)
anti_join(main, excludes)

#   id  name      city   zip
# 1  3 tamil manhattan 10019
# 2  1  mary  new york 10017

回答2:

Adjusting your second question (not tested)

test_dff %>%
  filter(!((contbr_nm == 'GAITHER, BARBARA' & contbr_city == 'APO AE' & contbr_zip == '9309') |
           (contbr_nm == 'PANIC, RADIVOJE' & contbr_city == 'PORSGRUNN' & contbr_zip == '3924') |
           (contbr_nm == 'KHAN, RAMYA' & contbr_city == 'NEW YORK' & contbr_zip == '2586') ))

回答3:

Here's an approach that creates a new variable by concatenating the values in the multiple columns you want to reference with your filter:

set.seed(15)
dfTest <- data.frame(matrix(round(rnorm(20),3), nrow=10))
dfTest$tempcol <- paste(dfTest$X1,dfTest$X2)

head(dfTest)
      X1     X2       tempcol
1  0.259  0.855   0.259 0.855
2  1.831 -0.365  1.831 -0.365
3 -0.340  0.166   -0.34 0.166
4  0.897 -1.243  0.897 -1.243
5  0.488  1.459   0.488 1.459
6 -1.255 -0.004 -1.255 -0.004

#Now remove the values by filtering on tempcol
dfTest %>%
  filter(tempcol != '0.259 0.855') %>%
  select(1:2) #omit tempcol in output

      X1     X2
1  1.831 -0.365
2 -0.340  0.166
3  0.897 -1.243
4  0.488  1.459
5 -1.255 -0.004
6  0.023 -0.021
7  1.091  0.032
8 -0.132 -1.167
9 -1.075 -0.520