统计与决策2025，Vol.41Issue(24)：35-40,6.DOI:10.13546/j.cnki.tjyjc.2025.24.006
大语言模型在统计学中的应用与展望
Application and Prospect of Large Language Models in Statistics
摘要
Abstract
This paper aims to discuss the application prospects of Large Language Models(LLMs)in statistics,in order to ad-dress the limitations of traditional statistical methods in processing unstructured text data.Through literature review and case anal-ysis,the internal relationship and influencing mechanism between LLMs and statistics are systematically sorted out.A statistical based framework for big language models is constructed,and its application advantages in data preprocessing,feature extraction,model training,prediction optimization,and other aspects are analyzed.The paper adopts a combination of theoretical deduction and empirical methods to reveal the path for LLMs to improve statistical modeling efficiency and accuracy through automated fea-ture extraction,cross modal data processing,and time series analysis.The results indicate that LLMs not only expand the applica-tion boundaries of statistics,but also promote innovation in fields such as statistical inference,anomaly detection,and decision support through the integration with deep learning,Bayesian inference,and other technologies.In the future,with the optimization of model size and the combination of interdisciplinary technologies,LLMs are expected to make breakthroughs in improving model interpretability,reducing computational costs,and enhancing robustness,further unleashing the value of data-driven deci-sion-making and injecting new momentum into the development of statistics in the field of artificial intelligence.关键词
大语言模型/统计学/自然语言处理/人工智能Key words
large language model/statistics/natural language processing/artificial intelligence分类
信息技术与安全科学引用本文复制引用
Yin Nan,Zhu Yingming..大语言模型在统计学中的应用与展望[J].统计与决策,2025,41(24):35-40,6.基金项目
国家自然科学基金面上项目(42471194) （42471194）
