Prediction and Analysis of Customer Complaints Using Machine Learning Techniques

Abstract

Businesses must prioritize customer complaints because they highlight critical areas where their products or services may be improved. The goal of this study is to use machine learning approaches to anticipate and evaluate customer complaint data. The current study used logistic regression and support vector machine (SVM) to predict customer complaints, and evaluated the datasets using machine learning techniques after collecting five distinct length datasets from the Consumer Financial Protection Bureau (CFPB) website and cleaning the data. Both logistic regression and SVM can accurately predict customer complaints, according to this study, but SVM gives the greatest accuracy. The current study also found that SVM provides the highest accuracy for a one-month dataset and Logistic regression provides for a three-month dataset. In addition, machine learning codes were utilized to display and tabulate consumer complaints across many dimensions.