tslib wrapper library for commonly used time series forecasting algorithms

tslib provides a common interface to build a variety of time series forecasting models with the following methods:

1. fit
2. forecast
3. residual_analysis
4. summary
5. plot_fit
6. get_fit_interval
7. get_forecast_interval

It supports the following models:

Model	Use Case
ARIMA,ARIMAX and SARIAMX	Common time series scenarios involving stable single seasonality and exogenous variables
Prophet	Time series with multiple seasonality (weekly, yearly) and shock effects due to special events: holidays etc
TBATS	Time series with multiple seasonality and sparse data
Croston, ADIDA, TBA, IMAPA	Handle time series with sparse data

The supported models are in ./tslib/models.py file.

For evaluation and validation following classes are present:

1. CrossValidation: Useful to produce model accuracy analysis over different training lengths and evaluation time horizons.
2. CumpinessEval: Useful to find out the extent of clumpiness in time series data and suggest the models to be used.

This is still a work in progress and I plan to include the following:

1. Package this repo into a python package installable via pip, conda or uv.

You can find out how to use this library by referring to ./examples folder

How to get started:

This project is still a work in progress, so there is no python package. For the time being you can start using this library by setting up a conda environment using the env.yaml file.

Run the following command:

conda env create -f env.yaml

This will install the dependencies. You can keep the tslib folder in your working directory and follow along the notebooks in examples directory.