Bi-Directional Causal Analysis using a Novel Coefficient of Causation
T. Dhilip *
Department of Civil Engineering, Osmania University, Hyderabad, Telangana, India.
Ramesh Chandra Bagadi
Department of Civil Engineering, Osmania University, Hyderabad, Telangana, India and Department of Civil Engineering, NSR Institute of Technology, Sontyam, Visakhapatnam, A.P, India.
M. Gopal Naik
Department of Civil Engineering, Osmania University, Hyderabad, Telangana, India.
Suresh Kumar N
Department of Civil Engineering, Osmania University, Hyderabad, Telangana, India.
*Author to whom correspondence should be addressed.
Abstract
The understanding and computation of both correlation and causation are of prime importance to infer relationships between the variables under consideration. Often made conclusions related to causation are subjective and are likely to vary across space and time. The current research investigation aims to develop a model for computing the causal impact of one variable upon another using a Novel Coefficient of Causation. The proposed coefficient involves computing the inner product of two specially transformed variables between which the coefficient of causation is computed. The variable sets considered for the analysis are the Anscombe’s Quartet datasets. It is found that the coefficient well explains the causal behavior between two variables and the same is illustrated through analysis carried out on the standard functions. The proposed model can effectively help in drawing the causal inference objectively. Further, the model also helps in understanding the strength and direction of the causation.
Keywords: Coefficient of causation, bi-directional causal analysis, anscombe’s quartet, MATLAB