CODEWITHSUNDEEP
rdplyr
Javier Fajardo
Javier Fajardo

Reputation: 797

Compare value in a column within groups in dplyr

I would like to compare the values inside a grouped data.frame using dplyr, and create a dummy variable, or something similar, indicating which is bigger. Couldn't figure it out!

Here is some reproducible code:

table <- structure(list(species = structure(c(1L, 1L, 1L, 2L, 2L, 2L), .Label = c("Adelophryne adiastola", 
"Adelophryne gutturosa"), class = "factor"), scenario = structure(c(3L, 
1L, 2L, 3L, 1L, 2L), .Label = c("future1", "future2", "present"
), class = "factor"), amount = c(5L, 3L, 2L, 50L, 60L, 40L)), .Names = c("species", 
"scenario", "amount"), class = "data.frame", row.names = c(NA, 
-6L))
> table
                species scenario amount
1 Adelophryne adiastola  present      5
2 Adelophryne adiastola  future1      3
3 Adelophryne adiastola  future2      2
4 Adelophryne gutturosa  present     50
5 Adelophryne gutturosa  future1     60
6 Adelophryne gutturosa  future2     40

I would group the df by species. I want to create a new column, can be increase_amount, where the amount in every "future" is compared to the "present". I could get 1 when the value has increased and 0 when it has decreased.

I have been trying with a for loop that goes throw each of the species, but the df contains over 50,000 of them and it takes too long for the times I will have to re-do the operation...

Someone know a way? Thanks a lot!

Upvotes: 3

Views: 6466

Answers (3)

kstew
kstew

Reputation: 1114

It sounds like you could use lag() to quickly find the difference over time. I would suggest restructuring your scenario (time) variable so that it can be intuitively reordered using R functions (i.e., arrange() will alphabetically reorder your scenario variable to future1, future2, present, which won't work in this case).

df <- data.frame(species=rep(letters,3),
                 scenario=rep(1:3,26),
                 amount=runif(78))
summary(df)
glimpse(df)
df %>% count(species,scenario)

df %>% 
  arrange(species,scenario) %>% # arrange scenario by ascending order
  group_by(species) %>% 
  mutate(diff1=amount-lag(amount), # calculate difference from time 1 -> 2, and time 2 -> 3
         diff2=amount-lag(amount,2)) # calculate difference from time 1 -> 3

The output from lag() will result in NA's for the first scenario values within each grouping, but the results can be easily changed using ifelse() statements or filter().

df %>% 
  arrange(species,scenario) %>% group_by(species) %>% 
  mutate(diff1=amount-lag(amount)) %>% 
  filter(diff1>0)

df %>% 
  arrange(species,scenario) %>% group_by(species) %>% 
  mutate(diff1=amount-lag(amount)) %>% 
  mutate(diff.incr=ifelse(diff1>0,'increase','no increase'))

Upvotes: 0

akrun
akrun

Reputation: 887991

We can do this with ave from base R 

table$increase_amount <-  with(table, as.integer(amount > ave(amount * 
         (scenario == "present"), species, FUN = function(x) x[x!=0])))
table$increase_amount
#[1] 0 0 0 0 1 0

Upvotes: 0

Scarabee
Scarabee

Reputation: 5714

You can do something like that:

table %>% 
  group_by(species) %>% 
  mutate(tmp = amount[scenario == "present"]) %>% 
  mutate(increase_amount = ifelse(amount > tmp, 1, 0))
# Source: local data frame [6 x 5]
# Groups: species [2]
# 
#                 species scenario amount   tmp increase_amount
#                  <fctr>   <fctr>   <int> <int>           <dbl>
# 1 Adelophryne adiastola  present      5     5               0
# 2 Adelophryne adiastola  future1      3     5               0
# 3 Adelophryne adiastola  future2      2     5               0
# 4 Adelophryne gutturosa  present     50    50               0
# 5 Adelophryne gutturosa  future1     60    50               1
# 6 Adelophryne gutturosa  future2     40    50               0

Upvotes: 5

Related Questions