This package contains an implementation of the the skew-t distribution of Azzalini (link). Its PDF is given by
where
Here we limit our attention to the 1-D version. A simplified description of the parameters goes like
- Location parameter (not necessarily the mean).
- Scale parameter (not necessarily the variance or standard deviation).
- Skewness parameter (not necessarily the skewness itself).
- Tailedness parameter (influences kurtosis).
After cloning the repository, use the following commands to install the necessary requirements:
pip install -r path/to/cloned/repo/requirements.txt
Then, run the demo script via
python -i tskew_demo.py
The generated images will be similar to
The -i flag makes the script run in interactive mode so that the plots do not immediately close on program termination. A simplified use-case might look like
from tskew.tskew import fit_tskew
loc_est, scale_est, df_est, skew_param_est = fit_tskew(my_data)