|
Activity Number:
|
445
- GOVT CSpeed 2
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Government Statistics Section
|
|
Abstract #318139
|
|
|
Title:
|
Quantile Regression for Classification in Large Data Sets Using the Alternating Direction Method of Multipliers Algorithm
|
|
Author(s):
|
Scott Samuel Coggeshall* and Xiao-Hua Zhou and Lan Wang
|
|
Companies:
|
VA Puget Sound Health Services Research and Development and Peking University and University of Miami Herbert Business School
|
|
Keywords:
|
quantile regression;
electronic health records;
statistical programming;
alternating direction method of multipliers
|
|
Abstract:
|
Quantile regression is a method for both prediction and classification with many appealing properties. However, traditional methods for fitting quantile regression models do not scale well when applied to datasets with large numbers of observations or predictors. We propose an algorithm for fitting penalized semiparametric quantile regression models based on Alternating Direction Method of Multipliers (ADMM). The ADMM algorithm allows for straightforward distributed model fitting, making the fitting of this model to large datasets feasible. We then demonstrate the method by developing a quantile regression-based prediction model for predicting which patients receiving care at Veterans Affairs medical centers will incur high healthcare costs in the coming year.
|
Authors who are presenting talks have a * after their name.
