Using Investor and News Sentiment in Tourism Stock Price Prediction based on XGBoost Model

The covid-19 has strongly hit the tourism industry, leading to downturn in tourism stock market. The bearish events in tourism industry can evoke public panic about future stock market. Tourism companies struggled to maintain their business and positively coped with the emergencies. Meanwhile, the influence of covid-19 was gradually weakened. As situation improved, some investors started to be optimistic about tourism stock prices in the long term. From the phenomenon, we reckon the investor sentiment can influence stock prices. Besides, news about finance and tourism can represent the industry situations and influence the stock prices. In our work, we used investor sentiment and news sentiment polarity combining history stock transaction data to build features in stock price prediction. Multinomial Naive Bayes was used to train and obtain amounts of investor comments sentiment. Lexicon based approach was used to obtain news sentiment polarity. We used XGBoost model to do stock price prediction based on the features we built. The results showed it's an effective way to use sentiment do tourism stock price since the sentiment attributes increased prediction precision. And we also found that XGBoost combined sentiment feature performed better than other algorithm such as SVM and ANN in prediction.