deepcharles / categorical-change-points

Change-point detection in categorical time series

A categorical signal $(y_{t})t$ is a discrete-valued sequence, meaning that $y{t}$ can take only $D$ different values. Computationnally, categorical signals are one-hot encoded: $y_t$ is represented by a multivariate binary signal $ (z_{t}){t} $ where $z{t} \in {0,1 }^D $ and $z_{t,d} = \mathbb{1}(y_t \text{ is equal to the d-th symbol})$.

Python

To install the Python package, run in the repository

cd python/
python -m pip install .

R

To install the R package...