Combining rows and generating category counts - r

Question

I want to be able to first combine rows with a similar attribute into one(for example, one row for each City/Year), and then find the specific counts for types of categories for each of those rows.
For example, with this as the original data:
City Year Type of Death
NYC 1995 Homicide
NYC 1996 Homicide
NYC 1996 Suicide
LA 1995 Suicide
LA 1995 Homicide
LA 1995 Suicide
I want to be able to produce something like this:
City Year n_Total n_Homicides n_Suicides
NYC 1995 1 1 0
NYC 1996 2 1 1
LA 1995 3 1 2
I've tried something like the below, but it only gives me the n_Total and doesn't take into account the splits for n_Homicides and n_Suicides:
library(dplyr)
total_deaths <- data %>%
group_by(city, year)%>%
summarize(n_Total= n())

Answer 1

You may do this
library(tidyverse, warn.conflicts = F)
df <- read.table(header = T, text = 'City Year TypeofDeath
NYC 1995 Homicide
NYC 1996 Homicide
NYC 1996 Suicide
LA 1995 Suicide
LA 1995 Homicide
LA 1995 Suicide')
df %>%
pivot_wider(names_from = TypeofDeath, values_fn = length, values_from = TypeofDeath, values_fill = 0, names_prefix = 'n_') %>%
mutate(n_total = rowSums(select(cur_data(), starts_with('n_'))))
#> # A tibble: 3 x 5
#> City Year n_Homicide n_Suicide n_total
#> <chr> <int> <int> <int> <dbl>
#> 1 NYC 1995 1 0 1
#> 2 NYC 1996 1 1 2
#> 3 LA 1995 1 2 3
Created on 2021-07-05 by the reprex package (v2.0.0)

Answer 2

If you don't have too many types of death, then something simple (albeit a little "manual") like this might have some appeal.
library(dplyr, warn.conflicts = FALSE)
df <- read.table(header = TRUE, text = 'City Year TypeofDeath
NYC 1995 Homicide
NYC 1996 Homicide
NYC 1996 Suicide
LA 1995 Suicide
LA 1995 Homicide
LA 1995 Suicide')
df %>%
group_by(City, Year) %>%
summarize(n_Total = n(),
n_Suicide = sum(TypeofDeath == "Suicide"),
n_Homicide = sum(TypeofDeath == "Homicide"))
#> `summarise()` has grouped output by 'City'. You can override using the `.groups` argument.
#> # A tibble: 3 x 5
#> # Groups: City [2]
#> City Year n_Total n_Suicide n_Homicide
#> <chr> <int> <int> <int> <int>
#> 1 LA 1995 3 2 1
#> 2 NYC 1995 1 0 1
#> 3 NYC 1996 2 1 1
Created on 2021-07-05 by the reprex package (v2.0.0)

Answer 3

You can first dummify your factor variable using the fastDummies package, than summarise(). This is a more general and versatile approach that can be used seamlessly with any number of unique types of death.
If you only have two types of death and will settle for a simpler (though more "manual") approach, you can use the other suggestions with summarise(x=..., y=..., total=n()
library(dplyr)
library(fastDummies)
df%>%fastDummies::dummy_cols('TypeofDeath', remove_selected_columns = TRUE)%>%
group_by(City, Year)%>%
summarise(across(contains('Type'), sum),
total_deaths=n())
# A tibble: 3 x 5
# Groups: City [2]
City Year TypeofDeath_Homicide TypeofDeath_Suicide total_deaths
<chr> <int> <int> <int> <int>
1 LA 1995 1 2 3
2 NYC 1995 1 0 1
3 NYC 1996 1 1 2