Fast Algorithms for Logistic Regression with Big Data

• Activity Number: 509
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): HaiYing Wang*+ and Rong Zhu and Ping Ma
• Companies: University of New Hampshire and Chinese Academy of Science and University of Georgia
• Keywords: Big data ; Classification ; Fast algorithm ; Logistic regression

Abstract

In this paper, we address the problem of calculating maximum likelihood estimates efficiently for logistic regression from large data. Two fast algorithms are introduced to solve problems induced by big data . We provide theoretical analysis based on approximated errors of the proposed fast algorithms, and compare the performance of our methods using synthetic and real data sets.