Online Program Home
My Program

Abstract Details

Activity Number: 186 - Contributed Poster Presentations: International Chinese Statistical Association
Type: Contributed
Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
Sponsor: International Chinese Statistical Association
Abstract #305220
Title: Detecting Statistical Interactions via Additive Neural Network
Author(s): Fan Wu* and Tianyang Hu
Companies: Purdue University and Purdue Statistics
Keywords: Interaction Detecting; Deep learning; Functional Relationship; Interpretation; Feature Selection; Additive models

Discovering statistical interactions is an essential step towards understanding the complex relationships among explanatory variables. We propose a new approach for detecting statistical interactions in data by fitting an additive neural network. Our method is based on comparing the performance between the full model and the restricted model, where the full model includes all potential variables while the restricted model excludes the pre-specified interaction terms. Our testing method targets at functional interactions and is free of distributional assumptions. In numerical experiments, we demonstrate that our method can accurately detect interactions under synthesize and real data.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program