A Hybrid Fuzzy-Based Classification Model for Consumer Behavior Prediction

Abstract

This study presents an FCBMA- Fuzzy Rule Base Concept through Bagging and Mapping Approach based analysis of customer behavior, considering big data. The data undergoes the preparation phase initially. Two crucial processes occur here: data cleaning. Normalized preprocessed data is designated as the training and testing datasets. The final extracted features are derived throuh bagging approach where each bag servs as an independent training set for classifiers employing an advanced fuzzy rule-based classification model. The integration of bagging-based classifiers from the training dataset with mapping functions from the testing dataset produces the final set of classifiers. The final prediction outcome is determined by reducing the classifiers. Ultimately, performance criteria including specificity, precision, accuracy, sensitivity, false positive rate (FPR), false negative rate (FNR), among others, are employed to evaluate the predictive model's performance against that of traditional classifiers.