Welcome to the second chapter of Time Series Modelling.
In the previous article we covered the basic assumptions of Time Series modelling and ways to check weather your series is stationary or not because time series modelling can only be done on a stationary series. We can also make a non-stationary series as a stationary series by various methods such as differencing , logging etc. which would be covered in later posts.
To read the previous article, Click on the below link.
In this article we cover the various components of a Time Series
- Secular Trend or Long-term Movement:
The tendency of a time series data to increase or decrease or stagnate during a long period of time is called the secular trend. For example an upward tendency in the population, sales of a product etc. or a downward tendency of deaths in country due to the advancement of technology.
2. Periodic Movement or Short term fluctuation
There are number of instances where we see the data repeats itself periodically over time therefore giving rise to the short term fluctuations in the data.
There are two types of short term fluctuations:
- Seasonal Variation: These are the variation in the data that have a regular and periodic manner over a span of less than a year. So, you can see the seasonal variation in the data if its recorded monthly, quarterly and so on.
- Cyclical Variation: These are the oscillatory movements in the time series data that happen in more than a year is called cyclical variation. They happen due to the ups and down recurring periodically (not necessarily uniform) example the business cycles that we see boom,recession. Depression, recovery.
3. Random or irregular variations:
These variations are the result of unforeseen and unpredictable forces which happen in an irregular manner, they do not show any definite pattern and there no regular period or time over which they occur. For Example increase in deaths by earthquake or an epidemic. Irregular variations are generally short-term but sometimes there effect is so intense, that they might give rise to cyclical movement.
Stay tuned for interesting updates on Time-series.