Deep Learning Models for Stock Market Prediction Using Optimization Approach

Sentiment analysis plays a significant role to make beneficial decisions about business investments in stock markets. Additionally, stock analysis can be used to determine a company's or organization's business. Due to its instability, a variety of political and economic circumstances, and changes in leadership, among other things, forecasting stock prices was more difficult. This prediction only loses effectiveness when used with historical or textual data. The prediction is more accurate when stock price information is combined with news sentiments. As a result, a prediction framework was created using sentiment analysis and both stock price and news sentiment data. From those stock data, moving average (MA), relative strength index (RSI)-dependent features and moving average convergence divergence (MACD), were initially extracted. To identify sentiments from those news data, conducted specific processes such as pre-processing, feature extraction as well as categorization. In the final categorization stage, to provide a precise sentiment prediction the utilized Deep Belief Network (DBN) weights will get tuned by Self Improved Whale Optimization Algorithm. A parametric and non-parametric analysis has been conducted for our proposed SIWOA by altering the parameters.