Trying to run many anovas and get an F value for each row

Question

I'm working with a dataset that looks like what is shown below. Now I know that this is not the type of format that R likes. I know how to tidy up the data, but then I'm not sure what I'd do in order to obtain an F statistic for each unique_id, which is my goal. Is there an easy way to do that? Otherwise is there a way I could use some type of apply function to tidy up each row independently, perform, the anova, and then add F statistics as a new column?

unique_id   heart   heart   heart   kidney  kidney  kidney  cortex  cortex  cortex
373020.8    1.39    1.18    1.30    2.71    2.96    2.52    1.97    1.67    1.44
371588.9    1.93    2.35    2.50    2.54    1.63    2.23    2.68    2.89    1.86
367772.8    0.42    0.51    0.97    1.02    0.03    0.82    0.01    0.90    1.01

How would you compute the F statistic for one row of this data frame? And is each row a unique unique ID? ie there's no repeats of `unique_id` in that column? — Spacedman, Feb 17 '17 at 18:19
Yes to your 2nd and 3rd question. for your first, i'd be comparing heart vs kidney vs cortex (each having three replicates) to get 1 F statistic for each row. I should have been more clear. — Eric Brenner, Feb 17 '17 at 18:26
What code precisely? Given a vector of nine values, grouped in threes like that, what would you run? — Spacedman, Feb 17 '17 at 18:32
By the way, data frames with repeated names are probably not a good thing... — Spacedman, Feb 17 '17 at 18:41

score 0 · Accepted Answer · answered Feb 17 '17 at 19:35

I'm partial to data.tables and you can easily do this with a DT after melting your data. Here's my take, with DT as your data.table containing the info you provided.

DT <- dt<- melt(dt, 
      id.vars = c("unique_id"),
      measure.vars = c("heart","cortex","kidney"))

DT[,fstat:=summary(aov(value~variable))[[1]][1,"F value"],by=unique_id]

This calculates the F-Stat by unique_id and should work.

Trying to run many anovas and get an F value for each row

1 Answers1

Linked