13 KiB
Chapter 3: Vector Autoregressive Methods¶
Previously, we have introduced the classical approaches in forecasting single/univariate time series like the Autoregressive-Moving-Average (ARIMA) model and the simple linear regression model. We learned that stationarity is a condition that is necessary when using ARIMA while this need not be imposed when using the linear regression model. In this notebook, we extend the forecasting problem to a more generalized framework where we deal with multivariate time series–time series which has more than one time-dependent variable. More specifically, we introduce vector autoregressive (VAR) models and show how they can be used in forecasting mutivariate time series.
The notebook is outlined as follows:
Multivariate Time Series model
Motivation
Univariate VS Multivariate Time Series
Examples
Foundations
Vector Autoregressive (VAR) Models
VAR(1) model
VAR(p) model
Choosing the order p
Building a VAR model
Structural Analysis
Impulse Response Function
Forecast Error Variance Decomposition
Takeaways
References
By students of PhD in Data Science Batch 2023 at the Asian Institute of Management
© Copyright 2020.