| In recent years,China’s pharmaceutical industry has been constantly pioneering and innovating,and the pharmaceutical industry is in a golden stage of rapid development,but with it,pharmaceutical safety issues and disturbances have also increased.The occurrence of major emergencies not only has an impact on the prospects of affiliated companies and industries,but even affects the country’s macroeconomics to a certain extent,especially the pharmaceutical industry issues related to people’s livelihood.These impacts are reflected in the stock market,manifested as fluctuations in stock prices of individual companies and industry sectors,and online search data as a reflection of people’s real-life behavior trends include investor attention to emergencies and dynamic changes in the stock market.This article takes the "Changchun Changsheng Biological Vaccine Counterfeit Incident",which has a great impact on the pharmaceutical industry in 2018,as the research object.The unexpected nature of the incident has caused a great negative impact and received widespread attention from the society.Produce violent shock.First,use the event analysis method based on web search data to explore the impact mechanism of emergencies on the stock price changes of the pharmaceutical industry sector;then predict the research on the trend of the comprehensive stock price index of the pharmaceutical industry,use the traditional time series analysis method to establish the AR model,and explore.The influence of online search data reflecting investor’s attention on stock price prediction is analyzed by comparing SVR algorithm models with different kernel functions and GBDT algorithm models with multiple loss functions.Finally,the empirical results show that major emergencies have a negative effect on the stock price of the pharmaceutical industry in the short term,and the cumulative abnormal yield in the event window is significantly affected;by comparing the prediction effects of the AR model,SVR and GBDT algorithm.After analysis,it was found that the SVR model added to the network search data has the highest prediction accuracy for the comprehensive stock index of the pharmaceutical industry in an emergency environment.The network search data can reflect the impact of the emergency on the stock market fluctuation of the pharmaceutical industry in a timely manner and improve the model’s prediction precision. |
