CODEWITHSUNDEEP
rdplyr
Moe
Moe

Reputation: 131

R - dplyr - filter top_n rows based on multiple conditions

Hi, this is my first post,

I hope to get it right and reproducible.

I was wondering if there is a more elegant solution than my approach below

I have a dataframe and would like to use conditional filters and extract rows that meet these conditions.

As output I would like the top_n rows that meet the conditional criteria (different conditions for top_n output from different columns), whilst preserving all other columns.

Example dataframe:

 set.seed(123)
  df1 <- data.frame(
  A = as.numeric(1:10),
  B = sample(seq(as.Date('2000/01/01'), as.Date('2018/01/01'), by="day"), size=10),
  C = as.numeric(sample(20:90, size = 10)),
  D = sample(c("yes", "no"), size=10, replace = TRUE),
  E = as.numeric(sample(1000:2000, size = 10))
)


df1 #check output

> df1 #check output
    A          B  C   D    E
1   1 2005-03-06 87  no 1963
2   2 2014-03-11 51  no 1902
3   3 2007-05-12 66  no 1690
4   4 2015-11-22 58  no 1793
5   5 2016-12-02 26  no 1024
6   6 2000-10-26 79  no 1475
7   7 2009-07-01 35  no 1754
8   8 2016-01-19 22  no 1215
9   9 2009-11-30 40 yes 1315
10 10 2008-03-17 85 yes 1229

Conditions I would like to use for filtering:

A) if column E is between 1000 and 1500 return top 2 rows weighted on column A

B) if column E is between 1000 and 2000 return top 2 rows weighted on column B

C) if column E is between 1000 and 1400 return top 2 rows weighted on column C

I have come up with the following solution but it is cumbersome and I wondered if there is a better approach. 

library("dplyr")
library("tidyr")
A<- df1 %>% dplyr::filter(E >= 1000 & E <= 1500) %>% top_n( n = 2, wt = A)  %>% arrange(-A) %>% mutate(condition = "-cond_A")
B<- df1 %>% dplyr::filter(E >= 1000 & E <= 2000) %>% top_n( n = 2, wt = B) %>% arrange(B) %>% mutate(condition = "cond_B")
C<- df1 %>% dplyr::filter(E >= 1000 & E <= 1400) %>% top_n( n = 2, wt = C) %>% arrange(-C) %>% mutate(condition = "-cond_C")

my desired output is the following:

spread(as.data.frame(distinct(bind_rows(A,B,C))),condition, condition)

   A          B  C   D    E -cond_A -cond_C cond_B
1  5 2016-12-02 26  no 1024    <NA>    <NA> cond_B
2  8 2016-01-19 22  no 1215    <NA>    <NA> cond_B
3  9 2009-11-30 40 yes 1315 -cond_A -cond_C   <NA>
4 10 2008-03-17 85 yes 1229 -cond_A -cond_C   <NA>

Would be great if you could tell me a better approach!

Upvotes: 3

Views: 2939

Answers (2)

Moe
Moe

Reputation: 131

that's great, thank you so much!

In my comments I asked if you could have more arguments to map2, and I realised that pmap can do just that.

pmap(list(c(1500, 2000, 1400), c(1000, 1700, 1300), names(df1)[1:3]),
     ~ df1 %>%
       filter(E >= ..2 & E <= ..1) %>%
       top_n(n=2, wt = !! rlang::sym(..3)) %>% 
       arrange_at(..3, funs(desc(.))) %>%
       mutate(condition = paste0("-cond", ..3)))  %>%
  bind_rows %>% 
  distinct %>% 
  spread(condition, condition)

Upvotes: 2

akrun
akrun

Reputation: 887991

We could use map2 from purrr to loop through the <= condition which changes and also the wt argument that takes the column names (based on the OP's code)

library(purrr)
library(dplyr)
library(tidyr)
map2(c(1500, 2000, 1400), names(df1)[1:3],
          ~ df1 %>%
                 filter(E >= 1000 & E <= .x) %>%
                 top_n(n=2, wt = !! rlang::sym(.y)) %>% 
                 arrange_at(.y, funs(desc(.))) %>%
                 mutate(condition = paste0("-cond", .y)))  %>%
                 bind_rows %>% 
                 distinct %>% 
                 spread(condition, condition) 
#   A          B  C   D    E -condA -condB -condC
#1  5 2016-12-02 26  no 1024   <NA> -condB   <NA>
#2  8 2016-01-19 22  no 1215   <NA> -condB   <NA>
#3  9 2009-11-30 40 yes 1315 -condA   <NA> -condC
#4 10 2008-03-17 85 yes 1229 -condA   <NA> -condC

Upvotes: 1

Related Questions