4 more MV Time Series Forecasting we should know — Auto_ARIMA, SARIMAX, VARMAX & Prophet

Image: source

In my earlier blogs, I’ve have discussed about multivariate time series forecasting using XGBoost Regressor and Vector AutoRegressor. In this blog we’ll see four more algorithms!

Datasets

Target Dataset

We want to forecast the future values of Production and Sales data for UK (source: OECD Statistics) assuming this is dependent on Consumer Price Index, Inflation Rate, Real Household Gross Disposable Income Per Capita and Population.

Target or endogenous variable to be predicted for the highlighted period.

Independent Variables

The independent variables are individually univariate-forecasted (for more details refer)

Training data

Around 3 years of actuals i.e. actual independent and dependent variables.