Presentation information

International Session

International Session » ES-2 Machine learning

[2S4-IS-2b] Machine learning

Wed. Jun 15, 2022 1:20 PM - 2:40 PM Room S (Online S)

Chair: Ken Ishibashi (University of Hyogo)

2:20 PM - 2:40 PM

[2S4-IS-2b-04] Product Portfolio Optimization for LTV Maximization

〇Kazuhiro Koike1, Noritomo Miyazawa1, Kenichi Machida1, Masumi Kawamura1, Kazuaki Takenaka1, Daishi Sagawa2, Kenji Tanaka2 (1. ASKUL Corporation, 2. Univ. of Tokyo)


Keywords:Modern portfolio theory, optimization problem, Mean variance model

To maximize LTV, it is common for internet shopping to conduct sales promotion periodically. Incentives such as coupons and points are given or attractive products are recommended in order to stimulate the purchase motivation. To maximize the effect with limited cost, it is important to narrow down the target to effective customers. We formulated this problem as a product portfolio optimization problem with the objective of maximizing LTV using Markowitz's mean-variance model. This model is generally used for deciding the portfolio of diversified investments of stocks. It can be applied to indices such as LTV, sales, and inventory cost in EC logistics. Demand fluctuation lead to the increase of cost, so the mean-variance model, which considers the fluctuation as a risk, good match for logistics. In this study, we constructed a mean-variance model for multiple indices of EC logistics, and related the models with a vector variable of product inclusion ratio to determine a recommended product portfolio that reduces the risk of fluctuation and achieves the expected LTV. As a result of verification with actual data, we confirmed the effectiveness of the proposed method.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.