Forecasting the impact of COVID-19 on GDP based on Adaboost

ABSTRACT

Covid-19 Pandemic has a unique impact on the economy as other infectious diseases. Epidemics affect people's daily consumption activities, for example, by causing them to shop less, travel less, consume less and invest less. The reduction of a large number of economic activities leads to the suppression of social demand and the reduction of consumption level, which further affects the GDP of various countries around the world. It is necessary to investigate and analyze the impact of the epidemic on GDP in order to control and analyze the economic situation under the impact of the epidemic. In this paper, we take the impact of COVID-19 on the GDP of each country as a regression problem, and propose to forecast GDP through feature engineering combined with Aaboost model. The model was tested on more than 50, 000 data records from more than 200 countries provided by the Kaggle platform to prove the validity. The experiment shows that Adaboost has stronger robustness compared with other methods, such as random forest, SVR. Adaboost improves the MSE of random forest by 2.39 and SVR by 0.38. © 2021 IEEE.