从 R 中的单个交易日期列中提取购买和销售日期

Question

I'm trying to split a transaction date column into 2 separate ones, 1 column for buy date, another for sell date. Likewise, I'd like to split a single transaction price column into sell price, and buy price. There is this post which is similar, but here I'd like to track every transaction date, instead of just imputing the earliest date as the buy and the latest date as the sell. So for example below is the current dataframe:

property = c('A','A','A','A','B','B','B')
transaction_dates = c("2011-03-09", "2013-06-06", "2015-08-28", "2016-07-18", "2016-12-13", "2018-10-29", "2019-11-30")
prices = c(750000, 830000, 820000,800000,825000,900000,600000) 

proptx = data.frame(property,transaction_dates,prices)

  property transaction_dates  prices
1        A        2011-03-09  750000
2        A        2013-06-06  830000
3        A        2015-08-28  820000
4        A        2016-07-18  800000
5        B        2016-12-13  825000
6        B        2018-10-29  900000
7        B        2019-11-30  600000

I am trying to add columns (or rather perhaps generate a new data frame) that breaks the transaction date column and price column into separate "buy" and "sell" columns, like so

  property    buy_date    buy_price  sell_date   sell_price
1        A    2011-03-09  750000     2013-06-06  830000
2        A    2013-06-06  830000     2015-08-28  820000
3        A    2015-08-28  820000     2016-07-18  800000
4        A    2016-07-18  800000     NA          NA
5        B    2016-12-13  825000     2018-10-29  900000
5        B    2018-10-29  900000     2019-11-30  600000
6        B    2019-11-30  600000     NA          NA

Ultimately what I would like to do is track the length of time that elapses between buy and sell dates, and then calculate the return to the seller. Rows 4 and 6 would represent that the property is not being/has not been sold. The actual data frame has hundreds of thousands of distinct properties, and I was hoping to do this sort of operation on each property.

Can this be done relatively easily?

Answer 1

Using data.table :

library(data.table)
dt <- as.data.table(proptx)
setnames(dt, old="transaction_dates", new="buy_date")
dt[, sell_date:=shift(buy_date, 1, type='lead'), by=property]
dt[, sell_price:=shift(prices, 1, type='lead'), by=property]
dt

   property   buy_date prices  sell_date sell_price
1:        A 2011-03-09 750000 2013-06-06     830000
2:        A 2013-06-06 830000 2015-08-28     820000
3:        A 2015-08-28 820000 2016-07-18     800000
4:        A 2016-07-18 800000       <NA>         NA
5:        B 2016-12-13 825000 2018-10-29     900000
6:        B 2018-10-29 900000 2019-11-30     600000
7:        B 2019-11-30 600000       <NA>         NA