MACHINE LEARNING MODELS FOR EXTRAPOLATIVE ANALYTICS AS A PANACEA FOR BUSINESS INTELLIGENCE DECISIONS

Abstract

The application of business intelligence (BI) in data analytics helps organizations access critical information in finance, marketing, healthcare, retail, and other critical infrastructures. However, there is a dearth of strategies to effectively leverage BI to empower businesses to refine useful data, understand newer industry trends, and improve competitive intelligence strategy for effective decision-making. This study implemented predictive data analytics to determine how the subjective decision-making process of used dealerships conducts their sales of vehicles and other business variable decisions. Scouring over forty-five different aspects of typical vehicle items, the study randomly selected twelve (12) features considered important. The data points were classified on the machine learning algorithms using a Support Vector Machine (SVM) to find the hyperplane of the (N-dimensional) features number for the training supervision of the dataset, while the Contingency Table Method (CTM) summarizes the relationship between the variables in the frequency distribution table. When six variables were outlined for comparison in the frequency distribution table, The models with optimal hyper-parameters showed similar predictive performances for all predictions while the “support vector regression algorithm” performs best with a strong output of 85% prediction analytics at a specific time of when certified used vehicles would be sold within a specified period. Consequentially, the extrapolative accuracy of the traditional decision-making process, when compared showed relative statistics of just around 50%. The study concludes that implementing business intelligence (BI) using machine learning models for predictive data analytics leads to increased revenue, effective customer satisfaction, an increase in market share, and improved decision-making.