A Novel Stock Price Prediction Model based on User Sentiments by Assimilating Bag of Words and Bi-Directional Gated Recurrent Unit

Stock price prediction based on user/ people sentiments relies on expressions, emojis, and statements posted in social media handles or blogs. A group of sentiments/ opinions impact the stock prices that need accurate prediction for safe investments. This article assimilates the Bag of Words (BOW) and Bi-Directional Gated Recurrent Unit (Bi-GRU) for reducing the mean average precision error in stock market price prediction. The BOW identifies the occurrences of the sentiment elements, such as words or emojis, and the recurrent unit categorizes them based on forward and backward sequences. By integrating these sequences, the highest match between the sequences and the emotions is identified to maximize the price forecast precision and error reduction. The proposed model reduces MAPE by 9.13 emotion recognition is the process of accurately interpreting