# Forecasting time series of binary vectors of given length

Please help me. I have a time series $$x^{(1)}, x^{(2)}, \ldots, x^{(n)}$$ where $$x^{(i)}=(x_1^{(i)}, x_2^{(i)}, \ldots, x_m^{(i)})$$, $$x_j^{(i)} \in \{0,1\}$$, that is every $$x^{(i)}$$ is a binary vector of fixed length $$m$$.

And I have to predict the next element $$x^{(n+1)}$$ of this time series.

As I see, it is the $$m$$-dimensional (multivariate) time series but its components are categorical (binary) variables.

What model is the most suitable for this task? What tool do I need to use for it?

• What's the problem in using some libraries such as prophet in python to do this work? – OmG Dec 3 at 9:27