ABSTRACT:
The recent advance of deep learning has enabled trading algorithms to predict stock value developments all the more accurately. Unfortunately, there is a critical gap in the real-world deployment of this breakthrough. This project presents DeepClue, a system develop to connect text-based deep learning models and end clients through visually interpreting the key elements learned in the stock price prediction model. We make three contributions to DeepClue. First, by designing the deep neural system design for interpretation and applying an algorithm to extract relevant predictive elements, we give a valuable case on what can be interpreted out of the prediction model for end clients. Second, by exploring hierarchies over the separated factors and showing these variables in an interactive, hierarchical visualization interface, we shed light on the most proficient method to successfully communicate the interpreted model to end clients. Third, we assess the integrated visualization system through two case studies in predicting the stock cost with online financial news and company related tweets from internet-based life. Quantitative examinations comparing the proposed neural system design and best in class models and the human baseline are conducted and reported. All the study results exhibit the effectiveness of DeepClue in helping to complete stock market investment and analysis tasks.