Stock Market Prediction using DNN
Abstract:- Stock Market Prediction using DNN –The stock market is a complicated and turbulent environment. Positive and negative attitudes based on media releases have an impact on it. The capacity to recognize stock movements determines the extent of stock price analysis. It is based on technical foundations and comprehension of the market’s hidden tendencies. Stock price prediction has always been a highly active topic of investigation and inquiry. However, achieving the appropriate level of precision remains a fascinating task.
In this research, we propose combining efficient machine learning approaches with a deep learning technique called Long Short Term Memory (LSTM) to accurately anticipate stock prices. Users’ emotions resulting from news headlines have a significant impact on traders’ buying and selling behaviors since they are quickly swayed by what they read. As a result, combining another dimension of sentiment with technical analysis should increase prediction accuracy. LSTM networks have been shown to be an effective tool for learning and predicting temporal data with long-term dependencies. In our research, the LSTM model is used to construct a better predictive model by combining historical stock data with feelings from news items.
Prediction of the Financial Markets Financial market forecasting has emerged as a new hot study issue in the machine learning industry in recent decades. Researchers overcome various challenges and made significant progress with the help of strong models like SVM (Support Vector Machine), feed-forward neural network, and recurrent neural network. Our thesis was influenced by their findings. We’ll go over a few of the most notable ones in this section.
Proposed model of the system:-
The data is supplied into the neural network, which is then trained for prediction with random biases and weights. Our LSTM model is made up of a sequential input layer, two LSTM layers, a dense layer with ReLU activation, and lastly a dense output layer with a linear activation function. The error, or difference between the target and the achieved output value, is reduced using a backpropagation technique that modifies the network’s weights and biases.
- Windows 7,8,10 64 bit
- RAM 4GB
- Python 2.7 or 3.x / Anaconda Navigator