Merging overlapp indices

I have a data.frame with start and end indices (sorted), for example:

df <- data.frame(start.idx = c(1,2,5),
                 end.idx = c(2,3,6))

I’m looking for a function that will merge rows i and i-1 if start.idx[i] == end.idx[i-1], such that the new row’s start.idx will be start.idx[i-1] and end.idx will be end.idx[i].

For the example above the resulting new (merged) data.frame will be:

res.df <- data.frame(start.idx = c(1,5),
                     end.idx = c(3,6))

>Solution :

You may create groups which include rows in the same group if there is a overlap and then select the first and last value for start and end respectively.

library(dplyr)

df %>%
  arrange(start.idx) %>%
  group_by(group = cumsum(start.idx > lag(end.idx, default = 0))) %>%
  summarise(start.idx = first(start.idx), 
            end.idx = last(end.idx)) %>%
  select(-group)


#  start.idx end.idx
#      <dbl>   <dbl>
#1         1       3
#2         5       6