摘要
为了识别出共享电动汽车使用意愿的关键影响因素,通过文献梳理总结了绩效期望、努力期望、社会影响、便利条件、价格价值、习惯、环保意识、政府政策、感知风险九个潜变量,运用因子分析和Lasso回归以及机器学习中特征提取的方法对527份调查问卷进行了分析。结果表明:综合传统量化研究的变量显著性、Lasso回归的正则化方法和机器学习模型的特征重要性度量,可以发现共享电动汽车的关键影响因素依次为环保意识、社会影响、价格价值、绩效期望和政府政策。研究结论对于新常态下如何推动我国电动汽车发展战略有重要的指导意义,同时将传统统计回归方法和机器学习的特征提取相结合的研究方法,也提供了用多方法相互验证来应对统计显著性的假阳性等问题的新思路。
To identify the determinants influencing the willingness to share electric vehicles,nine latent variables are summarized after literature review,including performance expectation,effort expectation,social impact,convenience,price value,habit,environmental awareness,government policy,and perceived risk;527 questionnaires are analyzed by factor analysis,Lasso regression,and the feature extraction method in machine learning.Combining the variable salience of traditional quantitative research,the regularization method of Lasso regression,and the feature importance measure of machine learning model,the research finds that the key factors influencing the sharing of electric vehicles are,in turn,environmental awareness,social impact,price value,performance expectation,and government policy.The research conclusion is of great significance to the development strategy of electric vehicles in China under the new normal state;meanwhile,the research method that combines the traditional statistical regression method with the feature extraction of machine learning provides a new way to deal with the false positives of statistical significance by using multiple methods to verify each other.
作者
朱振涛
杜明阳
刘颖
ZHU Zhentao;DU Mingyang;LIU Ying
出处
《江汉大学学报(社会科学版)》
2020年第6期91-102,127,共13页
Journal of Jianghan University(Social Science Edition)
基金
国家自然科学基金项目“考虑内生信息的城市交通合作演化机理研究”(71471084)
国家自然科学基金项目“互联网环境下考虑内生信息的邻避集群行为演化机理研究”(71571099)
江苏省高校哲学社会科学研究项目“基于协同消费理论的共享电动汽车创新推广机制研究”(2018SJA0395)
江苏省高校哲学社会科学优秀创新团队建设项目(批准号:2017ZSTD025)。