電子商務中基於大數據與機器學習的消費者分析： 探索行為導向性與促銷敏感度

Abstract

本研究探討了電子商務環境中消費者的線上購物行為類別與預測，旨在彌補傳統實體購物與線上購物在個人化服務方面的差距。透過現代電商平台提供的大量消費者點擊流數據，本研究開發了一套量化指標來分析消費者的線上瀏覽行為，從而詳細描繪出消費者的線上瀏覽行為軌跡以及行為模式。 本研究識別了兩個核心維度：搜索行為的導向性以及對促銷的敏感度，並將消費者劃分為四個獨特的行為類別：目標導向-促銷愛好者、目標導向-非價格驅動者、探索型-促銷愛好者、探索型-非價格驅動者。透過應用 K-means 分群演算法，驗證上述四種消費者行為類別的顯著性，不僅揭示了消費者購物行為的多樣性，也為電商平台提供了針對性的行銷策略方向。 為了在消費者購物初期階段提早識別出消費者的行為模式，本研究進一步開發了基於決策樹的機器學習模型。此模型能夠在早期瀏覽階段準確預測上述四個行為類別：目標導向-促銷愛好者、目標導向-非價格驅動者、探索型-促銷愛好者、探索型-非價格驅動者。這使得電商平台能夠及時且有效地實施針對性的市場策略。包括支持精準的廣告與促銷投放，實現推薦系統的客制化優化，並顯著提升個人化與互動性的使用者體驗。 總體而言，本研究不僅加深了對電子商務消費者行為的理解，還展示了利用數據驅動分析來提高商業決策效率的巨大潛力。這些發現為電商平台如何利用大數據及機器學習技術來優化行銷策略及顧客體驗提供了具體而實用的指導。

This study explores consumer online shopping behavior categories and predictions within the e-commerce environment, aiming to bridge the gap between traditional physical shopping and online shopping in terms of personalized service. Utilizing extensive consumer clickstream data provided by modern e-commerce platforms, this research developed a set of quantitative indicators to analyze consumers' online browsing behaviors, thereby detailing their browsing trajectories and behavioral patterns. The study identified two core dimensions: the Exploring Behavioral Orientation and Promotion Sensitivity, categorizing consumers into four unique groups: Goal-Oriented Promotion Lovers, Goal-Oriented Non-Price Driven, Exploratory Promotion Lovers, and Exploratory Non-Price Driven. By applying the K-means clustering algorithm, the significance of these four consumer behavior categories was validated, revealing the diversity of consumer shopping behaviors and providing targeted marketing strategy directions for e-commerce platforms. To identify consumer behavior patterns early in the shopping process, we further developed a decision-tree-based machine learning model. This model can accurately predict the aforementioned four behavior categories during the early browsing stage, enabling e-commerce platforms to implement targeted market strategies in a timely and effective manner. This includes supporting precise advertising and promotional campaigns, achieving customized optimization of recommendation systems, and significantly enhancing personalized and interactive user experiences. Overall, this research not only deepens the understanding of consumer behavior in e-commerce but also demonstrates the immense potential of data-driven analysis to enhance business decision-making efficiency. These findings provide concrete and practical guidance on how e-commerce platforms can leverage big data and machine learning technologies to optimize marketing strategies and customer experiences.