Predicting the direction of e-commerce stock prices during COVID 19 using machine learning

Abstract EN

The study investigates the predictive power of COVID 19 on e-commerce stocks in the United States and China.

technical and COVID 19 indicators were used as input features to predict the stock prices trend using three machine learning classifiers RF, SVM and KNN.

the performance of the classifiers is compared before and after performing feature selection technique with random forest feature importance on the dataset.

the results indicate that COVID 19 indicators have a predictive power on stock prices movements in both the US and Chinese markets.

the accuracy and F1 score metrics improve when using feature selection.

the performance of the classifiers with the selected features shows that RF outperforms the other classifiers with the highest accuracy and F1 score followed by SVM and KNN respectively.