R code optimization with a data.frame

Question

I have a Large SpatialPointDataFrame with 10570 elements, in which each row is a point with an associated date(some rows have the same date). This object has 4760 columns (it's the output of the extract() function between a RasterStack and the points) and each column corresponds to a date with an associated value (temperature).

Simplified example:

DATE2       BICHO   X2000.01.01   X2000.01.02   (...)   X2012.12.31
2009-04-08  Woody      20.7          19.2        ...         9.5
2009-04-09  Woody      20.7          19.2        ...         9.5
2009-04-10  Woody      20.7          19.2        ...         9.5
2004-11-30  Woody      20.7          19.2        ...         9.5
2004-12-01  Buzz       20.7          19.2        ...         9.5
2004-12-02  Buzz       20.7          19.2        ...         9.5

What I want to do is to create a new column (TP) in this data.frame, that contains the temperature for each corresponding date.

for(i in 11:4760){
  datas<-str_sub(colnames(pts@data[i]), start=2,end=11L)
  datas<-format(as.Date(datas, "%Y.%m.%d"),"%Y-%m-%d")
  for(j in seq_along(pts@data$TP)){
    print(c(i,j))   #just a print to see how fast is the code
    if(as.character(factor(pts$DATE2[j]))==datas){
      pts@data[j,]$TP<-pts@data[j,][i]
    }
  }
}

The code works but it's very slow, can anyone help me to optimize it?

Answer 1

There is no date in the columns that matches the DATE2 dates, but I hope this works for you:

library(data.table)
library(lubridate)
df = data.table(df) 
dfm = data.table:::melt.data.table(df,
                               id.var = c("DATE2","BICHO"), 
                               variable.name = "date", 
                               value.name = "TP")
dfm[,date := gsub("X","",date)]
dfm[,date := ymd(date)]
dfm[,DATE2 := ymd(DATE2)]
dfm[DATE2 == date,]