K-nearest neighbour methods for value of information in petroleum decision making

Abstract

In this thesis a parametric and non-parametric model are proposed for the study of value of information regarding decision making in petroleum industry. More specifically, a Gaussian method is introduced and analysed for the parametric model and k-nearest neighbour method is developed and discussed as a non-parametric representative. Both methods are implemented in R-code, a statistical programming language, and numerical results including qualitative plots and sensitivity tests are given. This paper concludes with a discussion of some important implementation concerns and recommendations for further work related to the topic.