1

I have one categorical variable with 12 classes (t) and one independent continuous variable (x1). One of the classes of the categorical variable is the reference class. I want to apply discriminant analysis to this bivariate data set. However, I cannot draw a graph because it has only one independent variable. How can I apply discriminant analysis to this data set, see class distinctions and draw graphics?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Buğra Varol
  • 11
  • 2
  • You can use the MclustDA procedure in R. – BigBendRegion Mar 05 '21 at 13:58
  • 2
    How many observations? Can you share with us a plot, maybe color-code the classes and plot it as a stripchart [one example here](https://stats.stackexchange.com/questions/93605/plotting-part-of-data-using-violin-plot-and-the-other-part-as-stripchart-stripp) with the 12 classes at different y plotting positions? Also consider share (a link to) the data! – kjetil b halvorsen Mar 05 '21 at 15:55
  • 1
    I'm not sure what the problem is with making a plot. Why not make some boxplots to start (`boxplot(x1~t)`)? There are *lots* of options here. Consider posting an example dataset. – gung - Reinstate Monica Mar 05 '21 at 19:18
  • Mathematically such discriminant analysis will be linearly equivalent (related) to one-way ANOVA. But it looks like the Q is not about DA, it is about plotting. – ttnphns Mar 05 '21 at 20:46

0 Answers0