Logarithmic Regret in Feature-based Dynamic Pricing

TL;DR

This paper introduces two algorithms for feature-based dynamic pricing that achieve optimal logarithmic regret bounds in both stochastic and adversarial settings, significantly improving upon previous results.

Contribution

It provides the first algorithms with optimal $O(d\,\log T)$ regret bounds for feature-based dynamic pricing in both stochastic and adversarial models.

Findings

Achieved $O(d\log T)$ regret bounds for stochastic and adversarial settings.

Improved previous regret bounds from polynomial to logarithmic scale.

Proved an $\\Omega(\sqrt{T})$ lower bound showing the benefit of demand curve knowledge.

Abstract

Feature-based dynamic pricing is an increasingly popular model of setting prices for highly differentiated products with applications in digital marketing, online sales, real estate and so on. The problem was formally studied as an online learning problem [Javanmard & Nazerzadeh, 2019] where a seller needs to propose prices on the fly for a sequence of $T$ products based on their features $x$ while having a small regret relative to the best -- "omniscient" -- pricing strategy she could have come up with in hindsight. We revisit this problem and provide two algorithms (EMLP and ONSP) for stochastic and adversarial feature settings, respectively, and prove the optimal $O(d\log{T})$ regret bounds for both. In comparison, the best existing results are $O\left(\min\left\{\frac{1}{\lambda_{\min}^2}\log{T}, \sqrt{T}\right\}\right)$ and $O(T^{2/3})$ respectively, with $\lambda_{\min}$ being the smallest eigenvalue of $\mathbb{E}[xx^T]$ that could be arbitrarily close to $0$. We also prove an $\Omega(\sqrt{T})$ information-theoretic lower bound for a slightly more general setting, which demonstrates that "knowing-the-demand-curve" leads to an exponential improvement in feature-based dynamic pricing.