Comparative Analysis of Machine Learning Techniques Using Customer Feedback Reviews of Oil and Gas Companies

Abstract

Sentiment analysis is the process of computationally identifying and categorizing opinions from a piece of text to determine whether the writer's attitude towards a practical topic, products or services is positive, negative or neutral. In this study, Machine Learning techniques are used to perform sentiment analysis on Oil and Gas customer feedback data. We present a comparison of different classification algorithms used for opinion mining, including Support Vector Machine (SVM), Naïve Bayes (NB), Instance Based Learning (IB3), Random Forest (RF), Partial Decision trees (PART), and Logit Boost (LB). Many studies have been performed on sentiment analysis in different sectors, but research into Oil and Gas customer feedback has been limited. Therefore, we have targeted a pathless sector, namely the Petroleum sector, where companies express their opinions towards specific products or services. Waikato Environment for Knowledge Analysis (WEKA) is used for experimental results. The WEKA environment is open source software entailing a collection of machine learning algorithms to solve data mining problems. The main aim of this study is to evaluate the efficiency of the above mentioned classifiers in terms of Precision, Recall, F-Measure and Accuracy. The findings of the comparison analysis indicate that the Naïve-Bayes classifier gives the best Accuracy of all classifiers. A small dataset could be considered as a limitation to our study due to the difficulty of gaining more datasets at the time of the research. However, this research will play a vital role for researchers in making decisions about the algorithm that they are going to use to solve their data mining problems. © 2020 ACM.