Machine Learning–Based Consumer Behavior Prediction Models for E-Commerce Platforms: Enhancing Digital Financial Inclusion and Market Accessibility

Abstract

This study examines the growing problem that e-commerce platforms increasingly rely on machine learning based consumer behavior prediction models, yet limited empirical evidence explains whether these systems improve only commercial efficiency or also strengthen digital financial inclusion and market accessibility for consumers. The purpose of the study was therefore to investigate how machine learning based prediction models influence digital financial inclusion, market accessibility, and overall e-commerce platform performance within a quantitative, cross-sectional, case-based research design. Using purposive sampling, data were collected from 312 active online consumers drawn from cloud based and enterprise style e-commerce platform cases who had prior experience with personalized recommendations and digital payment systems. The key variables were machine learning based prediction models as the independent variable, digital financial inclusion and market accessibility as explanatory variables, and e-commerce platform performance as the dependent variable. Data were analyzed through descriptive statistics, reliability testing, correlation analysis, and multiple regression using SPSS. The findings showed consistently positive perceptions across all constructs, with mean scores of 3.94 for machine learning prediction models, 3.87 for digital financial inclusion, 4.01 for market accessibility, and 4.08 for platform performance. Reliability was strong, with Cronbach’s alpha ranging from 0.84 to 0.89 and 0.91 for the full instrument. Correlation analysis revealed significant positive relationships between machine learning prediction and digital financial inclusion (r = .61, p < .001), market accessibility (r = .66, p < .001), and platform performance (r = .72, p < .001). Regression results further showed that machine learning prediction significantly explained 37.2% of the variance in digital financial inclusion (β = .61, p < .001) and 43.6% of the variance in market accessibility (β = .66, p < .001), while the combined model explained 58.4% of the variance in platform performance. The study implies that predictive analytics in e-commerce should be designed not only for conversion and profitability, but also for inclusive access, payment readiness, and equitable participation in digital markets.