Time series analysis is used to analyse specific data collected through time. In a nutshell it predicts the outcome of a variable when seasonality (i.e. time) has significant influence. It has to be evaluating consistent time intervals otherwise it skews the data and analysis.

This type of analysis must abide by some rules. For example, how strong are the effects of seasonality, how close is the forecast to the observed data in testing, and does it make sense?

Use cases for it would be such things as analysing tourism in countries with high tourism (i.e. New Zealand, United Kingdom, Sri Lanka, and so on).

In alteryx it can be used to measure the impact of COVID-19 on expected tourism numbers.

If we take New Zealand tourism data and predict what it would have looked like if COVID-19 never happened in alteryx using the "ETS" method we end up having this kind of prognosis:

Which shows that the seasonality would still stay and the data would be cyclical, although this changes when we include the massive drop in New Zealand's tourism due to covid.

We see that the predictive modelling is assuming there would still be seasonality in New Zealand's tourism before it would get back on track and rise to the levels that it used to be.

It should be noted that in time series predictive analysis the most recent data influences the outcome of the predictive algorithm most.

The outcome of such analysis usually ends up in three outputs:

Seasonality: short term cyclical movement

Trend: long term, generally non cyclical movement

Error: general variance in the data

Thank you for reading :)