• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2021 Fiscal Year Final Research Report

Investigating the identifiability of machine learning and its application to consumer behavior analysis

Research Project

  • PDF
Project/Area Number 20K22125
Research Category

Grant-in-Aid for Research Activity Start-up

Allocation TypeMulti-year Fund
Review Section 0107:Economics, business administration, and related fields
Research InstitutionIshinomaki Senshu University

Principal Investigator

Sato Toshikuni  石巻専修大学, 経営学部, 助教 (10878804)

Project Period (FY) 2020-09-11 – 2022-03-31
Keywords識別性 / 潜在的ディリクレ配分法 / ニューラルネットワーク / 潜在変数モデル / マーケティング尺度 / 消費者心理 / 消費者行動
Outline of Final Research Achievements

This study aims to investigate the identifiability of machine learning models such as latent dirichlet allocation (LDA) and neural network to improve the stability of their parameter estimations in consumer behavior analysis. The results of this study provide mainly two contributions. (1) LDA: The stability and identifiability of LDA in specific situations were clarified with simulation-based parameter recovery experiments. The author proposed Markov Chain Monte Carlo algorithms for the Bayesian estimation of constrained LDAs and showed their stability. (2) Neural network (autoencoder): The author proposed constrained estimation methods for neural networks and indicated that they provide stable and explainable results compared with an unconstrained standard neural network. However, these two main results were obtained under limited assumptions, so that the author will explore the extension and scope of the applications in the future study.

Free Research Field

マーケティング・リサーチ

Academic Significance and Societal Importance of the Research Achievements

本研究の学術的な意義は、社会科学の領域で重視されている識別性の側面から機械学習の応用を議論した点であると考えられる。特にマーケティングや消費者行動分野において、消費者心理のような観測できない要因や不確かな要因を仮説的に測定したり、モデルで記述したりすることは重要な役割を果たしてきた。しかし、それらについて同一のデータと手法を用いても著しく異なる結果が得られる場合には、意思決定に多重の不確実性を与えることになる。このような文脈で本研究の社会的意義は、実社会で急速に利用が進んでいる機械学習について、社会科学での理論的背景も考慮しながらより頑健な応用方法を探索したことであると考えられる。

URL: 

Published: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi