How to reorder facet_wrap based on two variables

Question

I need to make a bar plot based on two variables (points and type) with fill.

Below is a minimal example, I would like to see the points ranking by guard points and ranking by points as guard or forward.

I tried ~reorder(names, -c(type, points)) but it doesn't work.

    name <- c("James Harden","James Harden","Lebron James","Lebron James","Lebron James","Kawhi Leonerd","Kawhi Leonerd","Klay Thompson","Steph Curry","Kevin Durant","Kevin Durant","Chris Paul","Chris Paul")
    team <- c("HOU","OKC","LAL","MIA","CLE","SAS","TOR","GSW","GSW","GSW","OKC","HOU","LAC")
    points <- c(2000,12000,2000,10000,20000,7000,2000,14000,20000,6000,18000,4000,14000)
    type <- c("G","G","F","G","F","G","G","G","G","F","F","G","G")
    nba <- data.frame(name,team,points,type)
    nba <- nba %>% arrange(desc(type))
    ggplot(nba, aes(x = type, y = points, fill = team)) +
      geom_bar(stat = 'identity', position = 'stack', color = 'black') +
      facet_wrap(~reorder(name,-points),  ncol = 1, strip.position = "top") +
      coord_flip() + theme_minimal() +
      labs(x = "players", y = "points", title = "Rank by points as Guard")

If it's ranked by points as guard, I would like to see Steph Curry ranks top, Chris Paul at second, James Harden and Klay tied at third, Lebron at fifth, Kawhi at sixth, and KD at the bottom.

If it's ranked by points as either guard or forward, I'd like to see Lebron at top, KD second, so on and so forth.

It's not random! look at the results of ```reorder(nba$name,-nba$points)``` — M--, Jun 04 '19 at 18:09
@M-M maybe I'm dull but it doesn't seem to change anything for me. And if I remembered correctly, you don't really need to attach the dataset name in ggplot — timxymo1225, Jun 04 '19 at 18:14
I meant to look at it outside of `ggplot` Levels are based on `-points` — M--, Jun 04 '19 at 18:17
@M-M Interesting, in what ways can I change that order though? — timxymo1225, Jun 04 '19 at 18:31
`facet_wrap(~reorder(name, -points, sum)`? You are using the average points per category (team and position) for each player, which isn't very informative. — Axeman, Jun 04 '19 at 18:37
@Axeman This answers one of the questions, thank you so much! any comments on ranking by points for a specific type? — timxymo1225, Jun 04 '19 at 18:47

score 1 · Accepted Answer · answered Jun 04 '19 at 19:03

You can sort it for points as guard by adding a helper column. Look below;

library(ggplot2)
library(dplyr)

nba %>% 
 mutate(guardpoints = points * (type=="G")) %>% 
  ggplot(aes(x = type, y = points, fill = team)) +
  geom_bar(stat = 'identity', position = 'stack', color = 'black') +
  facet_wrap(~reorder(name, -guardpoints, sum),  ncol = 1, strip.position = "top") +
  coord_flip() + theme_minimal() +
  labs(x = "players", y = "points", title = "Rank by points as Guard")

nba %>% 
  ggplot(aes(x = type, y = points, fill = team)) +
  geom_bar(stat = 'identity', position = 'stack', color = 'black') +
  facet_wrap(~reorder(name, -points, sum),  ncol = 1, strip.position = "top") +
  coord_flip() + theme_minimal() +
  labs(x = "players", y = "points", title = "Rank by points")

^{Created on 2019-06-04 by the reprex package (v0.3.0)}

How to reorder facet_wrap based on two variables

1 Answers1