Patterns and components
In time series, there are mainly four different patterns or components:
- Trend: A slow but significant change of the values over time
- Season: A change that is cyclical and has a period of less than one year
- Cycle: A change that is cyclical and has a period of longer than one year
- Random: A component that is random; the best model for purely random data is the mean, given that it has a distribution corresponding to the normal distribution
Thus, before we can analyze our data, it needs to be stationary, and for it to be stationary, we need to take care of the patterns: trend, season, and cycle. The analysis that you will perform will be on part of the time series that does not fit into any of these patterns, with the random component ...
Get Mastering Python Data Analysis now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
