linear interpolation based on look up tables in R (Finding the Age Lower bound and Upper bound)-CodePudding

I have two data sets, For example:

# First data:
Age<-c(2,2.1,2.2,3.4,3.5,4.2,4.7,4.8,5,5.6,NA, 5.9, NA)
df1<-data.frame(Age)
# Second data:
Age2<-seq(2,20,0.25)    
Mspline<-rnorm(73)
df2<-data.frame(Age2, Mspline)

For each value of age in the first data set, I need to find the right value of Mspline (linear interpolation) based on the second data set by the following formula:

Suppose your age in the first data set is 4.8, so the lower bound and upper bound of this age is 4.75 and 5 (these are the two closest values in Age2). Then, the value of Mspline for this age would be:*

Age2      Mspline     Sspline 
4.75    -0.0769      0.1592 
5       -0.0752      0.1535 
Mspline = -0.0769   ((4.8-4.75)/0.25)·(-0.0752 - (-0.0769)) = -0.0766

Mspline at your age= Mspline of lower bound of age  ((your age-lower bound age)/0.25)*(Mspline of lower bound of age)-(Mspline of upper bound of age)

I was wondeing how I can make this function in R?

CodePudding user response：

Here's a function that implements the formula as I understand it:

Age<-c(2,2.1,2.2,3.4,3.5,4.2,4.7,4.8,5,5.6,NA, 5.9, NA)

# Second data:
Age2<-seq(2,20,0.25)    
Mspline<-rnorm(73)

res <- lapply(1:length(Age), \(x){
  lwr_ind <- max(which(Age2 <= Age[x]))
  upr_ind <- min(which(Age2 >= Age[x]))
  data.frame(Age = Age[x], 
             Mspline = Mspline[lwr_ind]   ((Age[x]-Age2[lwr_ind])/0.25)*(Mspline[lwr_ind] - Mspline[upr_ind]))
})

res <- do.call(rbind, res)
res
#>    Age     Mspline
#> 1  2.0 -0.48510574
#> 2  2.1 -0.45222184
#> 3  2.2 -0.41933793
#> 4  3.4 -1.05075284
#> 5  3.5  1.03415440
#> 6  4.2  0.90388824
#> 7  4.7  1.05441685
#> 8  4.8 -1.86696649
#> 9  5.0 -0.60582730
#> 10 5.6 -0.76802820
#> 11  NA          NA
#> 12 5.9  0.09647946
#> 13  NA          NA

^{Created on 2022-10-19 by the reprex package (v2.0.1)}

The remaining question is whether you're trying to do something with the missing values (NA). The missing values in Age continue to be missing in the result above.