A Study of Machine Learning-Based Methods for Analyzing Financial Statements

Abstract

Research proposes using machine learning for financial statement analysis to aid auditors facing large-scale financial data. Starting with suspicions around "high deposits and loans," Kangmei Pharmaceutical's audit concerns were scrutinized through financial statement analysis, A-share market analysis, and quantitative return rate analysis. The study established a hypothesis of significant audit risks and introduced two new audit approaches using big data: machine learning for income tax analysis in financial statements and time series analysis of Kangmei's stock market data. Findings indicate persistent instability in time series data, notably around a major misstatement in 2018. The return rate declined from 2014 to 2017 and spiked abnormally in 2018. These anomalies suggest abnormal fluctuations in Kangmei's A-share market data. In conclusion, leveraging big data technology can enhance audit efficiency and provide novel insights for auditors guided by audit standards ensuring thoroughness in critical assessments.