Goal is to produce a visualization indicating ratio.
library(tidyverse)
# Dataset creation
df <- data.frame(cls = c(rep("A",4),rep("B",4)),
grd = c("A1",rep("A2",3),rep(c("B1","B2"), 2)),
typ = c(rep("m",2),rep("o",2),"m","n",rep("p",2)),
pnts = c(rep(1:4,2)))
df
#### Data wrangling
df1 <- df %>%
group_by(cls) %>%
summarise(cls_pct = sum(pnts))
df1
df2 <- df %>%
group_by(cls,grd) %>%
summarize(grd_pct = sum(pnts))
df2
df3 <- df %>%
group_by(cls,grd,typ) %>%
summarise(typ_pct = sum(pnts))
df3
#### Attempt to combine all df1,df2,df3
# but mutate and summarise are mixing up leading to wrong results
df3 %>%
group_by(cls,grd) %>%
mutate(grd_pct = sum(typ_pct)) %>%
group_by(cls) %>%
mutate(cls_pct = sum(grd_pct))
joels
December 16, 2019, 6:34pm
2
Is this what you're trying to do:
library(tidyverse)
df %>%
group_by(cls,grd,typ) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts)) %>%
bind_rows(df %>%
group_by(cls, grd) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts),
typ = "All")) %>%
bind_rows(df %>%
group_by(cls) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts),
grd = "All",
typ = "All"))
cls grd typ pnts pct
<fct> <chr> <chr> <int> <dbl>
1 A A1 m 1 1
2 A A2 m 2 0.222
3 A A2 o 7 0.778
4 B B1 m 1 0.25
5 B B1 p 3 0.75
6 B B2 n 2 0.333
7 B B2 p 4 0.667
8 A A1 All 1 0.1
9 A A2 All 9 0.9
10 B B1 All 4 0.4
11 B B2 All 6 0.6
12 A All All 10 0.5
13 B All All 10 0.5
This can be shortened to the following:
groups = c("cls","grd","typ")
map_df(length(groups):1,
~ df %>%
group_by_at(groups[1:.x]) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts))) %>%
map_if(~!is.numeric(.), fct_explicit_na, "All") %>%
bind_cols()
And generalized with tidy evaluation:
my_summary = function(data, value.var, ...) {
groups = enquos(...)
map_df(length(groups):1,
~ data %>%
group_by_at(groups[1:.x]) %>%
summarise({{value.var}} := sum({{value.var}})) %>%
mutate(pct = {{value.var}}/sum({{value.var}}))) %>%
map_if(~!is.numeric(.), fct_explicit_na, "All") %>%
bind_cols()
}
my_summary(df, pnts, cls, grd, typ)
my_summary(diamonds, price, cut, color)
Thanks for so many solutions.
Can we use ggplot2 or sunburst or something to somehow show the proportion of 3 categories in 1 graph (e.g. bar-like chart) ? differentiating with colors or fill ? also, excluding "All" categories if possible.
please dont suggest facets as my real data has many variable
data <- df %>%
group_by(cls,grd,typ) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts)) %>%
bind_rows(df %>%
group_by(cls, grd) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts),
typ = "All")) %>%
bind_rows(df %>%
group_by(cls) %>%
summarise(pnts = sum(pnts)) %>%
mutate(pct = pnts/sum(pnts),
grd = "All",
typ = "All"))
#### Attempt to visualize all the ratios in 1 chart
data %>%
pivot_longer(cols = -c(cls:pnts),
names_to = "per_cat",
values_to = "percent") %>%
ggplot(aes(cls,percent, col = typ, fill = grd)) +
geom_bar(stat = "identity") +
coord_flip() +
theme_bw()
joels
December 16, 2019, 11:02pm
4
How about this:
library(tidyverse)
groups = c("cls","grd","typ")
1:length(groups) %>%
map_df(~ df %>%
group_by_at(groups[1:.x]) %>%
summarise(pnts = sum(pnts),
groups=unique(groups[.x])) %>%
ungroup %>%
mutate(pct = pnts/sum(pnts))
) %>%
mutate(subgroup = gsub("-NA", "", paste(!!!syms(groups), sep="-")),
groups = fct_rev(groups)) %>%
ggplot(aes(x=groups, y=pct, fill=subgroup)) +
geom_col(colour="white", width=0.98, show.legend=FALSE) +
geom_text(aes(label=paste0(gsub(".*-(.*)", "\\1", subgroup),
"\n",
sprintf("%1.1f", pct*100), "%")),
position=position_stack(vjust=0.5), colour="white", size=4) +
scale_fill_manual(values=rep("grey40", 13)) +
scale_y_continuous(expand=c(0,0), labels=scales::percent) +
theme_classic(base_size=15) +
theme(plot.margin=margin(r=15)) +
coord_flip() +
labs(x="Grouping Columns", y="")
Thanks a ton @joels , this is amazing.
