Applying mvn to iris requires subsetting by the Species variable.
However, the result of this package is a nested of named lists, finally containing a dataframe with p-values and metrics for each subset.
I find this uncomfortable to read and to process, so I want to merge all this almost equally named tables into one.
I haven't a found a merge function for multiple tables at once.
Therefore, I tried with Reduce() as suggested in other similar questions.
- The first applies a merge() correctly to pairs of dataframes.
- merging is done with merge() with 2 tables at the time.
The problems are:
- Naming of the classes (Species) are lost
- Results gets mixed
- I am forced to specify either univariate o multivariate, and call 2 times.
I would like just a resulting dataframe with, 1 column for the classes, 1 for the univariate/multivariate, and the rest merged by column names.
mvn_results = MVN::mvn( iris, subset='Species', mvnTest = "hz" )
mvn_results
Resulting:
$multivariateNormality
$multivariateNormality$setosa
$multivariateNormality$versicolor
$multivariateNormality$virginica
$univariateNormality
$univariateNormality$setosa
$univariateNormality$versicolor
$univariateNormality$virginica
$Descriptives
$Descriptives$setosa
$Descriptives$versicolor
$Descriptives$virginica
And tables with repeated structure like this:
I tried this:
mvn_merged <- Reduce(function(x, y)
merge( x, y, all=TRUE), mvn_results$univariateNormality )
mvn_merged
which produced the next result:
