Competitive Model Selection in Algorithmic Targeting

We study how market competition influences the algorithmic design choices of firms in the context of targeting. Firms face a general bias-variance trade-off when choosing the design of a supervised learning algorithm in terms of model complexity or the number of predictors to accommodate. Each firm has a data analyst who uses the chosen algorithm to estimate demand for multiple consumer segments, based on which it devises a targeting policy to maximize estimated profits. We show that competition induces firms to strategically choose simpler algorithms that involve more bias but lower variance. Therefore, more complex/flexible algorithms may have higher value for firms with greater monopoly power. History: Anthony Dukes served as the senior editor for this article. Funding: This work was supported by Hong Kong Research Grants Council [project number 14503122].

• 저널Marketing Science · 43(6) · 1226–1241
• 토픽Statistical Methods and Inference · Statistics and Probability
• DOI10.1287/mksc.2023.0175
• 저자Ganesh Iyer, T. Tony Ke