Design of stock price prediction model with various configuration of input features

The interest rate of Korea has remained around 1.75 % annually for years. However, the interest rates of developed countries such as the United States of America are 2.25 ~ 2.5%. Korea is very low compared to this, so it is hard to save money on a deposit or installment saving. Therefore, many people want to use stock investment methods to gain high interest rates despite the high risk. Many people are predicting whether stock prices will rise or fall for investments on their subjective opinion. However, in the field of computer engineering, many people try to predict the stock price using artificial neural network, which has been proven to have good performance through many studies. The direction of stock forecasting research using artificial neural networks is very diverse such as model structure, composition of input feature, composition of target vector and so on. In this paper, we design three stock price prediction model with various input features that have specific characteristic. We hypothesized that, for effective stock price prediction through artificial neural networks, using implicit meaning data. We also questioned which of the implicit data would be most predictive. To prove it, we suggest three stock price prediction model. We implemented these three models and experimented to performance evaluation. Through this, we find out what kind of features would be effective for stock price prediction.