compotime is a library for forecasting compositional time series. At the moment, it provides an implementation of the models described in the paper "Forecasting compositional time series: A state space approach" (Snyder, R.D. et al, 2017). It is constantly tested to be compatible with the major machine learning and statistics libraries within the Python ecosystem.
compotime is currently available for python 3.9, 3.10 and 3.11. It can be installed from PyPI:
pip install compotimeThis example uses adapted data on the popularity of programming languages (PYPL).
import pandas as pd
from compotime import LocalTrendForecaster, preprocess
URL = "https://raw.githubusercontent.com/mateuja/compotime/main/examples/data/proglangpop_sample.csv"
time_series = pd.read_csv(URL, converters={"Date": pd.Period}, index_col="Date").pipe(
preprocess.treat_small, 1e-3
)
model = LocalTrendForecaster().fit(time_series)
model.predict(horizon=10)For more details, see the Documentation.