dcast - error: Aggregate function missing

Question

A little background information regarding my question: I have run a trial with 2 different materials, using 2x2 settings. Each treatment was performed in duplo, resulting in a total number of 2x2x2x2 = 16 runs in my dataset. The dataset has the following headings, in which repetition is either 1 or 2 (as it was performed in duplo).

I would like to transform this into a dataframe/table which has the following headings, resulting in 8 columns:

This means that each treatment (combination of material, setting 1 and setting 2) is only shown once, and the parameter of interest is shown twice.

I have a dataset which looks as follows:

 code     rep material airspeed classifier_speed fine_fraction
1 L17       1 lupine         50             600          1
2 L19       2 lupine         50             600          6
3 L16       1 lupine         60             600          9
4 L22       2 lupine         60             600          12
5 L18       1 lupine         50             1200         4
6 L21       2 lupine         50             1200         6

I have melted it as follows:

melt1 <- melt(duplo_selection, id.vars = c("material", "airspeed", "classifier_speed", "rep"),
          measure.vars=c("fine_fraction"))

and then tried to cast it as follows:

cast <- dcast(melt1, material + airspeed + classifier_speed ~ variable, value.var = "value")

This gives the following message:

Aggregate function missing, defaulting to 'length'

and this dataframe, in which the parameter of interest is counted rather than both values being presented.

(1) Please don't paste data as an image, I'm not going to spend time transcribing it, especially when you have it in a convenient textual format on your console. It it much better to include sample data in your question by pasting the output from dput(x), where x is the sample (perhaps head(myframe)). (2) Is this data.table::dcast or reshape2::dcast? — r2evans
(3) The message about defaulting to 'length' is not an error, it is just a message telling you that you did not provide an aggregating function. If you read ?dcast (from whichever package), it mentions fun.aggregate=. — r2evans
I used reshape2::dcast My apologies for the picture, I will upload the data as a table in my question — Marjanne
It can be difficult to produce a good, reproducible question, and taking shortcuts frequently breaks things. In this case, Error: id variables not found in data: rep. There's an R package reprex that helps with this process, by ensuring that code you intend to post here does not error for (accidental, avoidable) reasons like this. — r2evans
We are still in the position that this is not an error, nor is it necessarily wrong. It is just informing you that widening/reshaping with dcast can result in data aggregation/combination, and you have not been explicit about how you want that done. You've not mentioned if the results from dcast(.) are correct; if they are, then that message goes away by adding dcast(., fun.aggregate=length). If not, then please edit your question and add what you expect the results to be. — r2evans

Marjanne Marjanne · Accepted Answer · 2021-04-30T06:55:57

Thanks for your effort and time to try to help me out, after a little puzzling I found out what I had to do.

I added replicate to each observation, being either a 1 or a 2, as the trial was performed in duplo.

Via the code

cast <- dcast(duplo_selection, material + airspeed + classifier_speed ~ replicate, value.var = "fine_fraction")

I came to the 5 x 8 table I was looking for.

dcast - error: Aggregate function missing

1 Answers