Data Analysis of the Factors Affecting Cryptocurrency by Using Machine Learning
摘要
The rapid development of financial technology brings convenience to our society but may also create risks. Based on the e-SQ (electronic service quality) framework, this study aims to investigate how cryptocurrency's ease of use, reliability, responsiveness, and security affect customer satisfaction in Japan. The number of respondents is 1967. The methods of data analysis are SHAP (SHapley Additive ExPlanations), a cooperative game theoretic approach, and Random Forest using Lasso Regression, which enables feature selection. It was found that ease of use and reliability impacted customer satisfaction. It is noteworthy that security did not affect customer satisfaction. Blockchain technology enhances the security of cryptocurrency. Two-factor authentication and Cold Wallet were introduced as the security measurements of cryptocurrency in the questionnaire. The result of the data analysis implied that the customers possibly have little interest in such security measurements in Japan. The factors of reliability were the reasonableness of fees and validity of the spread, which is the difference between bid and offer prices. Putting all this together, the results suggested that the customers were less concerned with risk and more concerned with benefits in Japan.