CODEWITHSUNDEEP
rdataframe
Dan Chaltiel
Dan Chaltiel

Reputation: 8494

Calculate comething based on several columns of data frame

I have a dataframe named cin with repeated measures at different times.

Here is the head(cin) : 

  rat       t0     t30     t60     t90    t120    t180    t240
1   2  368.140 601.379 683.489 487.085 423.255 306.132 253.108
2   2   35.000  27.000  20.000  12.000   9.000   7.000   7.000
3   2  633.048 546.406 449.268 383.821 358.186 342.300 336.481
4   3 1044.160 738.000 442.000 297.000 234.000 232.000 235.000
5   3   74.000  78.000  67.000  54.000  46.000  41.000  40.000
6   3  211.181 142.615 160.104 163.867 181.148 170.053 189.865

I want to use the calculate the area under the curve, with a formula I saw on this SO post.

Here is my attempt :

times = c(0,30,60,90,120,180,240)
y = as.vector(t(cin[,2:8][1,])) #transposal of the first line -_-
auc_of_first_line = sum(diff(times[order(times)])*rollmean(y[order(times)],2))

This works for the first line (but with the ugly transposal), but I cant seem to find out how to generalize to cin$auc = sum...

How can I do so ?

Upvotes: 0

Views: 56

Answers (1)

Matt Jewett
Matt Jewett

Reputation: 3369

There are probably more efficient ways of doing this. But this might get you what you're looking for.

times = c(0,30,60,90,120,180,240)

# Transpose cin dataframe
cin <- as.data.frame(t(cin[,2:8]))

# Apply the calculation to each column and bind results to bottom of dataframe
cin <- rbind(cin, apply(cin, 2, function(x) sum(diff(times[order(times)])*rollmean(x,2))))

# Set rowname for calculation to "AUC"
row.names(cin)[length(row.names(cin))] <- "AUC"

# Transpose cin dataframe back to original format
cin <- as.data.frame(t(cin))

Upvotes: 1

Related Questions