Other questions using "table" in their title are actually using data frame.
I want to keep this strictly about table object.
Suppose I have tables with same structure that I want to find the average of.
For example:
test1 <- head(table(iris$Sepal.Length, iris$Species))
(test1 + test1 + test1) / 3
> (test1 + test1 + test1) / 3
setosa versicolor virginica
4.3 1 0 0
4.4 3 0 0
4.5 1 0 0
4.6 4 0 0
4.7 2 0 0
4.8 5 0 0
However, it cannot be done by:
> mean(c(test1,test1,test1))
[1] 0.8888889
> sum(c(test1,test1,test1)) / 3
[1] 16
Best approach I could find was to make the objects into a list of tables and use Reduce function:
Reduce(`+`, list(test1, test1, test1)) / 3
Is there more simpler way to do it without going back and forth using list object?
We may loop over the array in the 1st two dimensions and get the mean
apply(replicate(3, test1), 1:2, mean, na.rm = TRUE)
-output
setosa versicolor virginica
4.3 1 0 0
4.4 3 0 0
4.5 1 0 0
4.6 4 0 0
4.7 2 0 0
4.8 5 0 0
Or loop over a single dimension and get the rowMeans/colMeans
apply(replicate(3, test1), 2, rowMeans, na.rm = TRUE)
setosa versicolor virginica
4.3 1 0 0
4.4 3 0 0
4.5 1 0 0
4.6 4 0 0
4.7 2 0 0
4.8 5 0 0
Both these methods are better than the Reduce approach with + especially when there are missing values as na.rm argument is found in both mean and rowMeans/colMeans
NOTE: replicate is used to create an array by replicating the object 'test1' n times.
If the object is already a list of tables, then convert to array with simplify2array before applying the apply
apply(simplify2array(list(test1, test1, test1)), 1:2, mean, na.rm = TRUE)
setosa versicolor virginica
4.3 1 0 0
4.4 3 0 0
4.5 1 0 0
4.6 4 0 0
4.7 2 0 0
4.8 5 0 0