How to repeat calculation on a subset of a database - r

Question

I have a large data frame, with many transects and for each of those transect I want to calculate for each year the intercept of the cros (x) and the value (y). Then I want to know how the intercept changed over the different years. I know how to calculate the intercept, however I have a lot of transects and I have to repeat this a lot, and I would like to do that more automatically.
So this is how my data looks like:
df
transects year cros value
10 1996 11 -3
10 1996 12 5
10 2005 11 -9
10 2005 12 -3
10 2010 11 -8
10 2010 12 -8
11 1996 11 7
11 1996 12 -4
11 2005 11 -6
11 2005 12 9
11 2010 11 6
11 2010 12 17
12 1996 14 -16
12 1996 15 -17
12 2005 14 -18
12 2005 15 -11
12 2010 14 16
12 2010 15 7
So I made a function, to subset the dataset and to do some calculation with this subset.
Here is the code. I used lapply because I want the outcome of the code to put in a list. However it could be that lapply is not the right function for this problem.
transect <- c(10, 11, 12)
o <- lapply(1:length(transect), function(i) {
s101 <- subset(df, along == transect[[i+1]])
# I want to create a subset for every transect and with that subset I want to do multiple calculations.
# Dune volume
# This makes sure that I have an intercept, also if there is no value above the 3
AUC96<-0
AUC05<-0
AUC10<-0
# Here I calculate the intercept for the different years.
d96 <- subset(s101, (cros >= 3.00) & (year == 1996))
AUC96<-sintegral(d96$cros,d96$value)$int
lengthdune96 <- max(d96$value)-min(d96$value)
AUC962 <- lengthdune96*8.00
AUC96 <- AUC96 +AUC962
d05 <- subset(s101, (cros >= 3.00) & (year == 2005))
AUC05<-sintegral(d05$cros,d05$value)$int
lengthdune05 <- max(d05$alti)-min(d05$value)
AUC052 <- lengthdune05*8.00
AUC05 <- AUC05 +AUC052
d10 <- subset(s101, (cros >= 3.00) & (year == 2010))
AUC10<-sintegral(d10$cros,d10$value)$int
lengthdune10 <- max(d05$value)-min(d05$value)
AUC102 <- lengthdune10*8.00
AUC10 <- AUC10 +AUC102
# Here the difference between the years
dune96.05 <- AUC05-AUC96
dune05.10 <- AUC10-AUC05
c(transect[[i+1]], dune96.05, dune05.10)
})
out <- as.data.frame(do.call(rbind, o))
However when I try this I get this error:
`Error in approx(x, fx, n = 2 * n.pts + 1) :
need at least two non-NA values to interpolate`
This is the first time that I try to make such a function, so it could be that I am doing this totally wrong. I hope that you can help me.
EDIT:
So after I changed the answer a bit, because that did not totally worked out. However I still get a error message and I am really stuck. I also tried different ways to solve this questions, eg looking at the plyr package, however I still get the same error questions:
So this is how my code looked like:
test<-lapply(unique(df$transect),function(i){s101 <- subset(df,df$transect==i)
{
AUC96<-0
AUC05<-0
AUC10<-0
d96 <- subset(s101, (cros >= 3.00) & (year == 1996))
AUC96<-sintegral(d96$cros,d96$value)$int
lengthdune96 <- max(d96$value)-min(d96$value)
AUC962 <- lengthdune96*8.00
AUC96 <- AUC96 +AUC962
d05 <- subset(s101, (cros >= 3.00) & (year == 2005))
AUC05<-sintegral(d05$cros,d05$value)$int
lengthdune05 <- max(d05$alti)-min(d05$value)
AUC052 <- lengthdune05*8.00
AUC05 <- AUC05 +AUC052
d10 <- subset(s101, (cros >= 3.00) & (year == 2010))
AUC10<-sintegral(d10$cros,d10$value)$int
lengthdune10 <- max(d05$value)-min(d05$value)
AUC102 <- lengthdune10*8.00
AUC10 <- AUC10 +AUC102
dune96.05 <- AUC05-AUC96
dune05.10 <- AUC10-AUC05
}
c(i,dune96.05, dune05.10)
})
However I still get this error message:
`Error in approx(x, fx, n = 2 * n.pts + 1) :
need at least two non-NA values to interpolate`
I am not really sure what I am doing wrong, the function should work like this. I hope that somebody can help me.

Answer 1

I see two problems with your use of lapply. You index transect as a list (it's a vector) and do not pass it (nor df) as an argument to your function in lapply, hence no luck with the subsetting. Try something like this:
lapply(unique(df$transect),function(i,df){s101 <- subset(df,transect==i)
,...
c(i, dune96.05, dune05.10)
},df)