A Novel Model Developed for Forecasting Oilfield Production Using Multivariate Linear Regression Method

Abstract

In this paper, a multivariate linear regression model was developed for
predicting crude oil production volume in a group gathering facility
within the Niger delta area of Nigeria.
The dataset used was split randomly into two parts namely the training
and testing data set. This was done to ensure the model was not over
fitted. The model depends on four (4) independent variables: volume
correction factor, metered volume, metered factor and gross standard
volume for accurate predictions of net oil volume (dependent variable).
The model was compared with other existing models and was found to be
more accurate and has better performance in terms of root mean square
error and residuals. This novel model is suggested for use by the oilfield
managers to assist in decision making.