Auto-feed hyperparameter support vector regression prediction algorithm in handling missing values in oil and gas dataset

There are many types of equipment involved in the oil and gas industry. However, they have their useful lives and will degrade over time. This issue prompts to be solved using predictive analytics to predict the Remaining Useful Life (RUL) of equipment. In the historical data, however, there are missing values due to broken equipment sensors probes and different time rate sensors. This can significantly affect the prediction results and making it less accurate due to missing value and become a challenging issue. Missing values in datasets is a synonymous problem in data mining which could lead to an incomplete dataset, making inaccurate predictions results in machine learning prediction processes. This problem inspires the idea to develop a prediction algorithm to predict the missing values in the dataset, where Support vector regression (SVR) has been proposed as a prediction method to predict missing values in several academic types of researches. SVR however is inferior in accuracy and thus this paper discusses the usage of an optimized SVR with Evolved Bat Algorithm (EBA) to handle the missing value accurately with high execution time. The paper also presents the topic of missing values in the dataset, as well as compares the performance of the optimized SVR with the original SVR in terms of accuracy and execution time while handling missing values in a large dataset. The novel optimization-based artificial intelligence algorithm proposed in this paper implies an improved way to overcome a real engineering challenge i.e. handling missing values for better RUL prediction, hence bringing great opportunities for the domain area. © 2020, World Academy of Research in Science and Engineering. All rights reserved.