Although this may seem like a straightforward task to some, as a beginner in R it has been frustrating! The task is as follows. I have a table with the following columns:

colnames(gov_data)
 [1] "year"               "quarter"            "employed"          
 [4] "newhires"           "separations"        "jobscreated"       
 [7] "jobsdestroyed"      "state"              "mw"                
[10] "teen_wage"          "teen_pop"           "adult_wage"        
[13] "teen_share_working" "unemp_primemale"    "recession"         
[16] "period" 

Using state_list<-split(gov_data, gov_data$state) I now have a list of data.tables corresponding to each state. Within each of these data.tables, I want to order by date. Here is how I did that. If this is inefficient, I welcome your alternatives!

orderfun <- function (x) {
  x[order(period)]
}

lapply(state_list, orderfun)

I now want to add a column labeled "change_mw" which corresponds to the change in the "mw" column. I know how to do that to a single data.table. I would create a column that lags so its the value of "mw" in t-1 and then take the difference between those two columns: one_table[,`:=` (mw_t_minus_1 = shift(mw,n=1,type="lag"), change_mw = mw - mw_t_minus_1) ][, mw_t_minus_1 = NULL ] How can I do this across multiple data.tables in a list? Is it even possible to use the data.table [i,j,by] in this instance? How would you go about this task? Once again, your help is very much appreciated!

CodePudding user response：

Here is an example that does similar, I'd be able to get closer with proper demo data

library(data.table)
dtCars <- data.table(mtcars, keep.rownames=TRUE)

dtCars[order(hp), change:= hp-shift(hp), by=cyl]