More Like this

1. DanceSport and Power Values

   Cheng, Diana; Coughlin, Peter (January 2021, The Montana math enthusiast)

   DanceSport is a competitive form of ballroom dancing. At a DanceSport event, couples perform multiple dances in front of judges. This paper shows how a goal for a couple and the judges' evaluations of the couple's dance performances can be used to formulate a weighted simple game. We explain why couples and their coaches may consider a variety of goals. We also show how prominent power values can be used to measure the contributions of dance performances to achieving certain goals. As part of our analysis, we develop novel visual representations of the Banzhaf and Shapley-Shubik index profiles for different thresholds. In addition, we show that the "quota paradox" is relevant for DanceSport events. 

   more » « less
2. Sparse regression for extreme values

   https://doi.org/10.1214/21-EJS1937  

   Chang, Andersen; Wang, Minjie; Allen, Genevera I. (January 2021, Electronic Journal of Statistics)
3. Reputation Effects under Interdependent Values

   Harry PEI (September 2020, Econometrica)

   Joel Sobel (Ed.)
4. Optimal clustering with missing values

   https://doi.org/10.1186/s12859-019-2832-3  

   Boluki, Shahin; Zamani Dadaneh, Siamak; Qian, Xiaoning; Dougherty, Edward R. (June 2019, BMC Bioinformatics)
5. Feature Selection using e-values

   Majumdar, S; Chatterjee, S (July 2022, Proceedings of the 39th International Conference on Machine Learning)

   In the context of supervised parametric models, we introduce the concept of e-values. An e-value is a scalar quantity that represents the proximity of the sampling distribution of parameter estimates in a model trained on a subset of features to that of the model trained on all features (i.e. the full model). Under general conditions, a rank ordering of e-values separates models that contain all essential features from those that do not. The e-values are applicable to a wide range of parametric models. We use data depths and a fast resampling-based algorithm to implement a feature selection procedure using e-values, providing consistency results. For a p-dimensional feature space, this procedure requires fitting only the full model and evaluating p + 1 models, as opposed to the traditional requirement of fitting and evaluating 2^p models. Through experiments across several model settings and synthetic and real datasets, we establish that the e-values method as a promising general alternative to existing model-specific methods of feature selection 

   more » « less