Understanding time series data
The objective of a time series machine learning algorithm is to forecast values and effectively plan the use of resources, such as inventories, seasonal-demand equipment allocation, and agriculture production, for example.
As a regression model needs a statistically significant relationship between the variables, a time series model needs autocorrelated data to be useful for a predictive model. In the following figure, we can see that the regression model variables' relationship is tested by statistical methods such as f-statistics and p-value:
Figure 3.1 – A: Linear regression and B: Air passenger time series
Figure 3.1 shows the prediction model for four trimesters of years 11 and 12 from air passenger time series data from the past 10 years. To build a useful predictive model, the air passenger data from years 1 to 10 needs to autocorrelate. This means that each value is dependent on prior data. Looking at the...
