When dealing with time-series of categorical data, basic quantities like mean value or variance are undifined (How to find the mean of ["cat", "dog", "cat", "cat, "dog"] ?). Concepts like stationnarity or correlations also need to be revisited.

I often had to deal with such time-series and implemented various standard methods in the present small package: CategoricalTimeSeries.jl.
The main functions were tested by reproducing results from the corresponding litterature.

I will keep adding utilities when I discover new concepts. I hope it can be usefull to you as well!