ARIMA Library
A library for time series prediction using ARIMA and its variants
Overview
This project aims at developing a generic library for analysing and exploring time-series and training ARIMA models for predicting values. It comprises of time-series analysis like checking for stationarity and seasonality using various statistical tests. Time-series visualisations also hold a large part in determining the nature of data. Further, it includes optimisation of ARIMA models by searching for the best possible values of p, d and q, using various algorithms like grid-search etc.
This project encompasses an amalgamation of various types of ARIMA models like ARMA, SARIMA, VARIMA and ARIMAX.
Technology used
Python, Statsmodels, Numpy, Pandas, Matplotlib