Machine learning approaches to time series analysis

Everything we have explained so far in this chapter may give the impression that ARIMA modeling is the only approach to time series modeling available. This is certainly not true. The rapid development and success of machine learning techniques over the last two decades has inevitably meant that machine learning algorithms have been applied to time series datasets.

However, the field of machine learning algorithms applied to time series is probably comparable to or bigger than that of ARIMA modeling, and it is still a rapidly developing field. As with other branches of machine learning, it is a very applied field. There are not simple clear-cut mathematical concepts or principles that define the field of machine learning for time series analysis, like there are for ARIMA analysis of time series. Therefore, in this section, we focus on giving you a short review of how the field has developed and its current state. Since it is a...