A multivariate distance nonlinear causality test based on partial distance correlation: a machine learning application to energy futures
Germán G. Creamer
School of Business Stevens Institute of Technology 1 Castle Point on Hudson, Hoboken, NJ 07030
Abstract: This paper proposes a multivariate distance nonlinear causality test (MDNC) using the partial distance correlation in a time series framework. As an extension of the Brownian distance correlation, partial distance correlation calculates the distance correlation between random vectors X and Y controlling for a random vector Z. Our test can detect nonlinear lagged relationships between time series, and when integrated with machine learning methods, can improve the forecasting power. We apply our method as a feature selection procedure and combine with the support vector machine and random forests algorithms to study the forecast of the main energy financial time series (oil, coal, and natural gas futures). It shows a substantial improvement on the forecast of the fuel energy time series in comparison to the classical Granger causality method in time series.