Forecasting Performance of Random Walk with Drift and Feed Forward Neural Network Models
Augustine Pwasong, Saratha Sathasivam · International Journal of Intelligent Systems and Applications · 2015
In this study, linear and nonlinear methods were used to model forecasting performances on the daily crude oil production data of the Nigerian National Petroleum Corporation (NNPC).The linear model considered here is the random walk with drift, while the nonlinear model is the feed forward neural network model.The results indicate that nonlinear methods have better forecasting performance greater than linear methods based on the mean error square sense.The root mean square error (RMSE) and the mean absolute error (MAE) were applied to ascertain the assertion that nonlinear methods have better forecasting performance greater than linear methods.Autocorrelation functions emerging from the increment series, that is, log difference series and difference series of the daily crude oil production data of the NNPC indicates significant autocorrelations.As a result of the foregoing assertion we deduced that the daily crude oil production series of the NNPC is not firmly a random walk process.However, the original daily crude oil production series of the NNPC was considered to be a random walk with drift when we are not trying to forecast immediate values.The analysis for this study was simulated using MATLAB software, version 8.03.