How to summarise overlaps between data points

问题

I have a data set of animals passing an RFID reader, it looks like this -

ID    date_time                 
A     2019-11-02 08:07:47    
B     2019-11-02 08:07:48 
A     2019-11-02 08:07:49
A     2019-11-02 08:07:50
A     2019-11-02 08:09:12
A     2019-11-02 08:09:13
B     2019-11-02 08:09:17

I asked this question recently, (combine multiple rows into one time interval), and now my data looks like this - (with the data organised into ten second intervals)

ID     start_date_time.      end_date_time
A      2019-11-02 08:07:47   2019-11-02 08:07:50
B      2019-11-02 08:07:48   2019-11-02 08:07:48
A      2019-11-02 08:09:12   2019-11-02 08:09:13
B      2019-11-02 08:09:17   2019-11-02 08:09:47

I have also added a column which summarises the intervals

dat$Interval = interval(dat$start_date_time,dat$end_date_time)

I now need to find and summarise where these intervals intersect and produce this as a count, to show the number of times animals interact (or are present at the RFID reader at the same time) something like this - (and without repeating reverse interactions, i.e. A-B, B-A)

ID     ID2     Interactions(n)
A      A       0
A      B       1    
A      C       3

Any help appreciated.

回答1:

Not really easy to answer, this question. Maybe this might be a good starting point:

library(tidyverse)
library(lubridate)

cbind(
  dat[rep(1:nrow(dat[-1, ]), nrow(dat[-1, ]):1), c(1, 4)],          
  setNames(
    dat[unlist(sapply(2:nrow(dat), seq, to = nrow(dat))), c(1, 4)],
    c('ID2', 'Interval2')
    )
  ) %>%
  mutate(
    interacts = intersect(Interval, Interval2),
    Interval  = NULL,
    Interval2 = NULL
  ) %>%
  filter(!is.na(as.numeric(interacts))) %>%
  count(ID, ID2)

# # A tibble: 1 x 3
#   ID    ID2       n
#   <chr> <chr> <int>
# 1 A     B         1

来源：https://stackoverflow.com/questions/58957189/how-to-summarise-overlaps-between-data-points

标签

social-networking

lubridate

rfid