Calculating percentage change from a fixed year for different group

Question

I would like to calculate the percentage change of my variable "var2" for different cities over time relative to the year "2000" ?

I tried this:

library(dplyr)
data <- data.frame(cities= c('NY','NY','NY','NY','NY','PL','PL', 'PL','PL','PL','AS','AS','AS','AS','AS','RY','RY','RY','RY','RY', 'JK', 'JK', 'JK', 'JK', 'JK'), year=c('2000', '2002', '2004', '2006', '2008', '2000', '2002', '2004', '2006', '2008','2000', '2002', '2004', '2006', '2008','2000', '2002', '2004', '2006', '2008','2000', '2002', '2004', '2006', '2008'), 
                    var2 = c(12,26,17,8,14, 12,20,10,8,14,12,20,10,8,14,12,20,10,8,14,12,20,10,3,5))

changes <- data2 %>%
    group_by(cities) %>%
    arrange(year, .by_group = TRUE) %>%
    mutate(variable_change = round((var2/lag(var2) - 1)*100, digits = 1))

But it calculates the percentage change between each year and I'm trying to calculate the changes between 2000 and 2002, 2000 and 2004 and so on...

Answer 1

You can use match to get corresponding var2 where year = 2000 and divide it with var2 value in each city.

library(dplyr)

data %>%
  group_by(cities) %>%
  mutate(variable_change = var2/var2[match(2000, year)])

#  cities year   var2 variable_change
#   <chr>  <chr> <dbl>           <dbl>
# 1 NY     2000     12           1    
# 2 NY     2002     26           2.17 
# 3 NY     2004     17           1.42 
# 4 NY     2006      8           0.667
# 5 NY     2008     14           1.17 
# 6 PL     2000     12           1    
# 7 PL     2002     20           1.67 
# 8 PL     2004     10           0.833
# 9 PL     2006      8           0.667
#10 PL     2008     14           1.17 
# … with 15 more rows

We can use also use == if it is guaranteed to have only 1 year with value 2000 in each city.

data %>%
  group_by(cities) %>%
  mutate(variable_change = var2/var2[year == 2000])

Answer 2

We can use %in% and it would also work when there are NAs

library(dplyr)
data %>%
      group_by(cities) %>%
      mutate(variable_change = var2/var2[year %in% 2000])