نوع مقاله : علمی-پژوهشی
نویسندگان
1 دانشکده مهندسی کامپیوتر، دانشگاه یزد، یزد، ایران
2 عضو هیات علمی گروه کامپیوتر/دانشگاه یزد
3 عضو هیات علمی دانشگاه یزد
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
In this paper, the impact of sentiment analysis on increasing the accuracy of stock price predictions using technical indicators is investigated. Six basic estimators and five effective technical indicators are used to predict stock prices. The time window sizes were set at 3, 7, 14, 30, 45, and 60 days, respectively, and the best time window was selected to predict stock prices. The goal was to predict stock prices after 7 days. The best models for predicting stock prices, based on MSE assessment criteria, were linear regression models, multilayer perceptron regression, and random forest regression with MSE values of 34.78, 37.78, and 1%, respectively. The results showed that the volume of tweets was associated with the volume of trade, as well as the mean score of positive sentiment in one day, with stock prices.
کلیدواژهها [English]