A Data-Driven Newsvendor Problem : Insights From Machine Learning and Feature Data Analysis

Analytically solving the classic newsvendor problem relies on a known demand distribution. In practice, however, it is not easy to approximate a sufficiently accurate demand distribution. This results in the obtained analytical solution may not be optimal. To address this issue, we propose a novel data-driven newsvendor model via responsible artificial intelligence (AI) algorithms. Our designed AI (machine learning) algorithms can incorporate demand features into replenishment decisions for solving both single- and multi-product newsvendor problems. In addition, they do not require the demand distribution and only rely on historical data. With the extracted key features of demand information from historical data, our responsible AI algorithms can simultaneously optimize demand estimation and replenishment decisions. First, given historical demand data with high-dimensional and mixed-frequency features, we propose a frequency alignment method to transform high-dimensional mixed-frequency data and historical data into the same frequency; this can effectively avoid losing the demand information. Second, two variable selection policies are proposed based on the empirical risk minimization principle to tackle the parameter proliferation issue; we numerically justify that the optimal order quantity can be obtained with the minimum total cost. Third, a feature-based machine learning algorithm is designed for solving the extended multi-product newsvendor problem. We find that using real-time information to update high-dimensional mixed-frequency data can effectively reduce the total cost. Numerical experiments validate that the identified key features of market demand can be sufficiently exploited to make data-informed replenishment decisions. We show that the optimal replenishment quantity behaves robustly and does not fluctuate greatly when given different cost structures. Our work provides meaningful insights for newsvendors making multi-product replenishment decisions under stochastic market demand. First, responsible AI can facilitate a human retailer and machine collaboration and provide a retailer with a competitive advantage by sharping and unpacking historical value. Second, the feature-based machine-learning algorithms proposed in our work can provide timely analytics and data-driven insights to make better replenishment decisions