Time Series Analysis,
Time Series Forecasting, and
Time Series Modeling are widely used. But I don't see the clear difference in between.
Does time series analysis mean the statistical analysis of time series.
Do the time series modeling and forecasting deal with the machine learning methods of dealing with time series.
Can anyone explain the differences between these terms?
I think these are often used colloquially as synonyms, but let's try to find the differences.
Each of them begins with "Time Series" (TS). So the difference lies in the three following terms. here with my interpretation:
- Analysis - wanting to describe and understand characteristics the observed data coming from the generating function$^1$.
- Forecasting - wanting to somehow predict the future output of the generating function.
- Modeling - wanting to describe and understand the generating function itself, via an approximation.
So as you see, I think time-series analysis and time-series modelling are almost the same thing. I just feel the latter most likely involves really using a real model, with parameters. That doesn't mean specifically parametric or a non-parametric approaches.
I personally wouldn't draw a line between them in terms of "uses statistics" vs. "uses machine learning", as your question suggests.
A natural progression in approaching a time-series problem (in my opinion) would be:
- understand the data (TS-analysis)
- understand the underlying function (TS-modelling)
- test how well I have understood the function and its output (TS-forecasting)
$^1$ With the term "generating function", I don't mean the strict mathematical definition, rather just the underlying process or system that is generating the observed data