使用 base 或 dplyr 在几列上过滤 dataframe

Question

I'm trying to filter the rows of a data frame based on columns in another datagram. Basically, I want to extract rows with the same IDs where the position is between start and end. There is the extra trick that the IDs are formatted differently.
finally, the data involved in the script is huge so to save memory or speed is nice to have.
would be grateful to get some tips.

library(dplyr)

df1 <- data.frame(id = c(1, 1, 1, 2, 2, 2, 3, 3, 3), 
                  pos = c(30, 40, 50, 35, 45, 55, 60, 63, 39))

df2 <- data.frame(idstr = c("id1", "id1", "id3", "id4", "id4"), 
                  start=c(30, 20, 30, 40, 20 ),
                  end = c(40, 30, 50, 60, 45))

df.base <- df1[ paste0("id", df1$id) == df2$idstr && 
                 df1$pos >= df2$start &&
                 df1$pos <= df2$end,]

df.dplyr <- df1 %>%
            left_join(df2, by  = c('id' == 'idstr') ) %>%
            filter(pos >= start & pos <= end) %>%
            select(id, pos)

edit: expected output, the rows from df1 meeting the condition (their position is in a range of df2 with the same id), so if no mistake: id, pos
1, 30
1, 40
3, 39

explanation: for example, df1[3,] id == 1 and pos == 50 looking at df2, there is no row where df2$id == "id1" and df2$start <= 50 and df2$end >= 50, so df1[3,] would be filtered out.

Answer 1

We can use non-equi join in data.table . Create the 'id' similar in both datasetss and then join on the 'id' columns and non-equi join with 'pos' and 'start', 'end' columns

library(data.table)
setDT(df1)[, id := paste0('id', id)]
df1[df2, on = .(id = idstr, pos >= start, pos <= end)]

Answer 2

I have taken your 2 DF df1 and df2 , mutated column idstr from df2 into a numeric by extracting the digits. Then with a left_join , group_by and filter I get the result.

library(dplyr)


df1 <- data.frame(id = c(1, 1, 1, 2, 2, 2, 3, 3, 3), pos = c(30, 40, 50, 35, 45, 55, 60, 63, 39))

df2 <- data.frame(idstr = c("id1", "id1", "id3", "id4", "id4"), 
                  start=c(30, 20, 30, 40, 20 ),
                  end = c(40, 30, 50, 60, 45))


df2 %>% 
  mutate(idstr = as.numeric(stringr::str_extract(idstr, '[0-9]'))) %>% 
  left_join(df1, by = c('idstr' = 'id')) %>% 
  dplyr::filter(pos >= start & pos <= end)
#> # A tibble: 4 x 4
#> # Groups:   idstr [2]
#>   idstr start   end   pos
#>   <dbl> <dbl> <dbl> <dbl>
#> 1     1    30    40    30
#> 2     1    30    40    40
#> 3     1    20    30    30
#> 4     3    30    50    39

---

There is one df1$id == 1 which fits into 2 start-end slots in df2 . And therefore it has to be 3 positions with id =1. If one of the limits is exlusive - like in the following code - it fits your wish.

df2 %>% 
  mutate(idstr = as.numeric(stringr::str_extract(idstr, '[0-9]'))) %>% 
  left_join(df1, by = c('idstr' = 'id')) %>% 
  dplyr::filter(pos > start & pos <= end)

#>   idstr start end pos
#> 1     1    30  40  40
#> 2     1    20  30  30
#> 3     3    30  50  39