以Apache Spark分散式平台為基礎之新電子優惠券推薦

Abstract

電子優惠券能吸引顧客消費，並提昇電子商務效益，然而電子優惠券對於不同顧客有不一樣的吸引成效。因此進行個人化的電子優惠券推薦，推薦顧客感興趣的電子優惠券，以增加顧客消費，是重要的研究議題。 本研究以電子優惠券使用資料來進行大數據分析，並使用Apache Spark大數據分散式處理平台提供的機器學習法，來預測使用者感興趣之新電子優惠券。本研究透過建立延伸屬性、屬性選擇、資料抽樣及分散式運算，分別以支持向量機、梯度提升決策樹、以及隨機森林演算法進行預測模型建置。本研究評估比較各預測模型的成效，實驗結果顯示隨機森林演算法有最佳之預測效果。

E-coupons can allure customers to make purchase and increase profits for e-commerce. However, e-coupons may have different effects on attracting customers, since customers may have different interests in e-coupons. Therefore, it is an important issue to recommend customers interested e-coupons and thus increase customer purchases through personalized recommendations of e-coupons. This research analyzes the usage data of e-coupons to predict user interests in new e-coupons by using the machine learning methods provided on the Apache Spark platform. The steps of building derived attributes, feature selection, data sampling and distributed processing are carried out to build the prediction models by using the SVM (Support Vector Machines), GBTs (Gradient-Boosted Trees), Random Forests, respectively. This research evaluates and compares the effectiveness of the prediction models. The experiment results show that the Random Forests performs better than the other methods.