ARIMA models

Until the advent of the ideas underpinning ARIMA modeling, time series analysis largely lacked a rigorous foundation. A large part of the uptake, popularity, and hence success of ARIMA models is the rigorous foundations that have been developed. Those foundations were primarily developed by the famous statisticians George Box and Gwilym Jenkins in the late 1960s and 1970s.

ARIMA modeling provides us with a mathematical framework to generate auto-correlation in a time series using very simple equations that specify how the time series evolves. Because of its simplicity and power, ARIMA modeling has for many years been considered the classic and only way to approach time series modeling. Only recently have modern machine learning methods such as deep learning neural networks begun to rival these classic ARIMA methods. Therefore, even if your preference is for more modern techniques, there is still huge value in learning and understanding these classic methods.

The...