sarahromanes/multida

High Dimensional Discriminant Analysis in R :sparkles:

multida's Issues

Error if Missing Values

Is it feasible for it to work if there are missing values in the data set?

SRBCT$X[3, 3] <- NA
res  <- multiDA(SRBCT$X, SRBCT$y, penalty="EBIC", equal.var=TRUE, set.options="exhaustive")
Error in rbind(deparse.level, ...) : 
  numbers of columns of arguments do not match

sparsediscrim package's dlda function behaves similarly.

Better incorporate `ggplot2` functionality in `plot.multiDA'

Currently - no functionality is available with plot.multiDA to change the style to how the user pleases. Task: to follow guide from here for best practices in incorporating ggplot2 in other packages.

Data Frames Have Incompatible Dimensions

I have a data set with 2038 samples belonging to six classes. Classifier training fails because mR and add don't have the same number of columns.

> mR <- rbind(mR,add)
Error in rbind(deparse.level, ...) : 
  numbers of columns of arguments do not match

Steps to reproduce are:

1. Download, from Dropbox, the R Data file Single Cell Research/SCProj-ClassifyR/data/segerstolpe_forDario.RData

load("segerstolpe_forDario.RData")
library(multiDA)
multiDA(t(segerstolpe), factor(cellTypes))

Installation of multiDA

Hi Author of multiDA! Just had a read of the package multiDA. Looks amazing and thought to implement it. By any chance, do you know which version of R does the package support? as it doesn't support version R 3.5.2. Many thanks in advance.