Scaled Beta Models and Feature Dilution for Dynamic Ticket Pricing

  • 2025-07-31 17:55:07
  • Jonathan R. Landers
  • 0

Abstract

A novel approach is presented for identifying distinct signatures ofperforming acts in the secondary ticket resale market by analyzing dynamicpricing distributions. Using a newly curated, time series dataset from theSeatGeek API, we model ticket pricing distributions as scaled Betadistributions. This enables accurate parameter estimation from incompletestatistical data using a hybrid of quantile matching and the method of moments.Incorporating the estimated $\alpha$ and $\beta$ parameters into Random Forestclassifiers significantly improves pairwise artist classification accuracy,demonstrating the unique economic signatures in event pricing data.Additionally, we provide theoretical and empirical evidence that incorporatingzero-variance (constant-value) features into Random Forest models acts as animplicit regularizer, enhancing feature variety and robustness. Thisregularization promotes deeper, more varied trees in the ensemble, improvingthe bias-variance tradeoff and mitigating overfitting to dominant features.These findings are validated on both the new ticket pricing dataset and thestandard UCI ML handwritten digits dataset.

 

Quick Read (beta)

loading the full paper ...