|
|
|
|
from insight to action: a lightweight ai framework for customer value maximization in sme e-commerce for emerging economies
|
|
|
|
|
|
|
|
نویسنده
|
sabet amir
|
|
منبع
|
اولين همايش ملي هوش مصنوعي در اقتصادهاي نوظهور و كسب و كارهاي ديجيتال - 1404 - دوره : 1 - اولین همایش ملی هوش مصنوعی در اقتصادهای نوظهور و کسب و کارهای دیجیتال - کد همایش: 04250-25067 - صفحه:0 -0
|
|
چکیده
|
In the rapidly expanding digital landscape of emerging economies, small and medium-sized enterprises (smes) face a critical challenge: while rich in customer data, they often lack the financial resources and specialized expertise to deploy complex, enterprise-grade artificial intelligence systems for customer analytics. this paper aims to bridge this capability gap by presenting and validating “from insight to action,” a novel, lightweight, and integrated three-stage ai framework specifically designed for sme e-commerce businesses. the primary objective is to provide a practical, low-cost methodology that transforms raw transactional data into actionable strategies for maximizing customer lifetime value (clv). the framework’s methodology unfolds in three sequential stages. first, the segment stage employs rfm (recency, frequency, monetary) analysis combined with the k-means clustering algorithm to identify distinct, behavior-based customer groups. second, the predict stage utilizes a highly efficient light gbm model to accurately forecast the future purchasing behavior and clv for each customer. finally, the act stage translates these predictive insights into a clear action-decision matrix, prescribing tailored marketing interventions for each segment. a proof-of-concept was implemented and rigorously validated using the publicly available “online retail dataset.” the results demonstrate the framework’s practical utility: the segmentation model successfully identified four meaningful and strategically relevant customer personas: ‘champions,’ ‘loyal customers,’ ‘at-risk customers,’ and ‘lost.’ furthermore, the light gbm-based clv prediction model achieved a high degree of accuracy, evidenced by a mean absolute error (mae) of 48.72 and a root mean square error (rmse) of 95.45, confirming its reliability. this research confirms that the proposed end-to-end framework serves as an accessible and powerful tool for smes, enabling them to make data-driven decisions, enhance customer retention, and achieve sustainable growth without significant capital investment in complex infrastructure.
|
|
کلیدواژه
|
artificial intelligence ,emerging economies ,sme e-commerce ,customer lifetime value (clv) ,machine learning ,data-driven decision making
|
|
آدرس
|
, iran
|
|
پست الکترونیکی
|
amirsabet@outlook.com
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Authors
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|