Enhancing E-Commerce Predictive Analytics using Ensemble Models for Consumer Behaviour and Market Forecasting

Predictive analytics is becoming more and more crucial in the fast-changing e-commerce sector, which helps businesses predict consumer behavior, enhance marketing, and optimize operational efficiency. This project aims to improve predictive analytics in e-commerce using machine learning techniques. AI is a fast-developing field that plays a significant role in data analysis and forecasting, allowing businesses to make informed decisions. The research utilizes algorithmic learning that involves ensemble models such as Kernel Support Vector Machine, Kernel Principal Component Analysis, and a Mentor-Learner based Optimization model. The main aim is to understand the potential growth and prospects of online shopping in the future. To achieve this, secondary data on ecommerce valuation was analyzed to generate revenue insights and understand online customer behavior better. Then, Python was applied to predict the sector's contribution from 2024 to 2026. The study also incorporates factors such as the proportion of Indian online consumers and e-commerce revenue to improve market price prediction accuracy. By combining academic research with industry practices, this paper offers valuable recommendations for online shopping managers and executives seeking to implement neural networks for analytical forecasting. The MAPE for KPCA-SVM-MLBO was 0.189%, significantly lower than the MAPE values of 0.769% and 0.547% for PCA-SVM-PSO and SVM-MLBO, respectively.