نجمه شاطري

Financial markets are constantly changing and people have always been interested in stock predictions. One of the important characteristics of stocks is uncertainty and it is difficult to predict what will happen in the future and cause losses. The financial market is a dynamic system that is complex, evolving and non-linear. Data intensity, noise, non-stationarity, unstructuredness, high levels of uncertainty and hidden relationships are the characteristics of the field of financial forecasting. It is difficult to discover the stock law and it is not possible to explore it with traditional methods, which will take a lot of money and time. Stock market forecasters focus on developing approaches to accurately predict index values or stock prices with the goal of earning large returns using specific trading techniques. The key to accurate stock market forecasting is to produce the best results with the least possible inputs and the simplest possible stock market model. Due to the technological advances in computer hardware and advances in machine learning techniques in recent years, more creative methods have been proposed. By reviewing the literature and eva‎luating data mining models in the field of stock price prediction, it has been shown that this field of machine learning methods and neural networks works well on real data. One of the most suitable machine learning algorithms in this field is the short-term long-term memory network LSTM, which we have used in this research, in addition, it has been combined with CNN to achieve the best results and performance of this model. In this research, the data of Metal Group in Tehran Bahadur Stock Exchange has been used to eva‎luate the combined CNN-BiLSTM model. At the end, the performance of the selected model is eva‎luated by MAE and RMSE.

قيمت سهم , پيش‌بيني , الگوريتم‌هاي داده‌كاوي

Stock price , forecasting , data mining algorithms

Najme Shateri

Dr. Saeed Mirzamohammadi