The choice of the solution method in the construction of the forecast model
Today, let’s talk about choosing a solution method when building a forecast model. Consider the advantages and disadvantages of different methods.
- Models of forecasting time series of the SARIMA family (AR, MA, ARIMA, SARIMA)
Advantage: A widely used method for predicting time series
Disadvantage: Slow learning curve, requires large computational resources, inability to take into account the influence of external factors.
- Linear regression
Advantage: Fast learning curve
Disadvantage: The accuracy of the forecast, the inability to specify categories
- Random Forest Regression
Advantage: Ability to add external factors and categories, built-in methods for assessing the significance of individual characteristics, scalability
Disadvantage: Slower in learning than linear regression; increased memory requirement.