CODEWITHSUNDEEP
rdplyr
Jazzmatazz
Jazzmatazz

Reputation: 645

Is it possible to use filter within mutate?

In my example below, I'm trying to do a mutate that does a rank within filtered criteria. My end goal is to filter, but keep the full data frame. I simply want my ranks to be done based on the filtered criteria. My df_goal below gets me the correct ranks, but would require me to then use a left_join to add them to the original df which seems like a waste. Is there a solution within dplyr that would allow me to rank the original df with the filter criteria? I was thinking that perhaps a pipe within a mutate would work, but I had no success.

Example:

library(tidyverse)

df <- tibble::tribble(
        ~slugSeason,         ~namePlayer, ~groupPosition, ~slugPosition, ~ptsPerGame, ~astPerGame,
          "2019-20",      "Aaron Gordon",            "F",          "PF",        11.3,           3,
          "2019-20",     "Aaron Holiday",            "G",          "PG",           3,         1.8,
          "2019-20",       "Abdel Nader",            "F",          "SF",           3,         0.3,
          "2019-20", "Admiral Schofield",            "F",          "SF",         4.4,         0.2,
          "2019-20",   "Al-Farouq Aminu",            "F",          "PF",         4.2,           1,
          "2019-20",        "Al Horford",            "F",          "PF",        18.8,         4.2,
          "2019-20",        "Alec Burks",            "G",          "SG",         8.5,         1.8,
          "2019-20",       "Alex Caruso",            "G",          "SG",         3.2,         1.6,
          "2019-20",          "Alex Len",            "C",           "C",         4.2,         1.6,
          "2019-20",  "Alfonzo McKinnie",            "F",          "SF",           0,           0
        )

df <- df %>% 
  group_by(groupPosition) %>%
  mutate(ptsRankbyPosition = rank(desc(ptsPerGame))) %>%
  ungroup() 

df_goal <- df %>% 
  group_by(groupPosition) %>%
  mutate(ptsRankbyPosition = rank(desc(ptsPerGame))) %>%
  ungroup() %>% 
  filter(slugPosition %in% c("PG", "SG", "SF") & ptsRankbyPosition > 1 |
           slugPosition == "C" & ptsRankbyPosition > 1) %>% 
  mutate(addlRank = rank(desc(ptsPerGame)))
Created on 2019-11-05 by the reprex package (v0.3.0)

Thanks in advance.

Upvotes: 3

Views: 955

Answers (1)

akrun
akrun

Reputation: 887691

One option is using replace after creating a logical column

library(dplyr)
df %>% 
 group_by(groupPosition) %>%
 mutate(ptsRankbyPosition = rank(desc(ptsPerGame))) %>%
 ungroup() %>%
 mutate(index = slugPosition %in% c("PG", "SG", "SF") & 
           ptsRankbyPosition > 1 | 
          slugPosition == "C" & ptsRankbyPosition > 1,
        addlRank = NA_real_, 
        addlRank = replace(addlRank, index, rank(desc(ptsPerGame[index])))) %>%
 select(-index)
# A tibble: 10 x 8
#   slugSeason namePlayer        groupPosition slugPosition ptsPerGame astPerGame ptsRankbyPosition addlRank
#   <chr>      <chr>             <chr>         <chr>             <dbl>      <dbl>             <dbl>    <dbl>
# 1 2019-20    Aaron Gordon      F             PF                 11.3        3                   2     NA  
# 2 2019-20    Aaron Holiday     G             PG                  3          1.8                 3      3.5
# 3 2019-20    Abdel Nader       F             SF                  3          0.3                 5      3.5
# 4 2019-20    Admiral Schofield F             SF                  4.4        0.2                 3      1  
# 5 2019-20    Al-Farouq Aminu   F             PF                  4.2        1                   4     NA  
# 6 2019-20    Al Horford        F             PF                 18.8        4.2                 1     NA  
# 7 2019-20    Alec Burks        G             SG                  8.5        1.8                 1     NA  
# 8 2019-20    Alex Caruso       G             SG                  3.2        1.6                 2      2  
# 9 2019-20    Alex Len          C             C                   4.2        1.6                 1     NA  
#10 2019-20    Alfonzo McKinnie  F             SF                  0          0                   6      5

Upvotes: 2

Related Questions