R code : Creating lagged Xs and y for supervised learning

This post shows a simple R code to create various lagged time series and concatenate them with the original time series. This can be used frequently when preprocessing time series data for machine/deep learning models.

A humpback whale

Creating lagged Xs and y

Time series and its various lagged one are used as input variables for supervised machine or deep learning models. The following R code generates this concatenation of a set of lagged and time-t variables of the time series.

As an output format, Xs (lagged variables) are followed by y since it is relatively easy to access X by using "1:nk" rather than "2:(nk+1)".

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

#========================================================# # Quantitative ALM, Financial Econometrics & Derivatives  # ML/DL using R, Python, Tensorflow by Sang-Heon Lee  # # https://shleeai.blogspot.com #--------------------------------------------------------# # Create lagged Xs and y for supervised learning #========================================================#   graphics.off(); rm(list = ls())   #=============================================== # function for creating lagged Xs and y #===============================================   func_lagged_Xs_y <- function(y, k, nmx = "x"){       nk <- length(k); ny <- length(y)     df <- as.data.frame(matrix(nrow=ny, ncol=nk))     colnames(df) <- paste0(nmx,k); df$y = y          for(i in 1:nk)          df[(1+k[i]):ny,i] <- y[1:(ny-k[i])]          return(df[(max(k)+1):ny,]) }   #----------------------- # sample data #----------------------- y <- c(31.12, 27.95, 30.67, 27.18, 21.89, 19.90, 21.58,         18.69, 20.31, 21.89, 19.29, 20.57, 21.57, 22.87,         21.01, 18.63, 17.96, 17.68, 17.54, 16.95, 17.33)   #----------------------- # test cases #----------------------- Xy1 = func_lagged_Xs_y(y, k=1:10) Xy1   Xy2 = func_lagged_Xs_y(y, k=c(1,3,5), nmx = "XX") Xy2   Colored by Color Scripter

cs

As expected, the results of two examples in the above R code are as follows.