CART를 이용한 주가결정요인분석과 주가예측: 코스피 200을 중심으로

Existing stock price analysis studies tended to focus only on stock price rather than stock price determinants, but in this study we develop a stock price forecasting system with a two-step method. In the first step, we find the factors that determine the stock price and predict the stock price based on the factors determined in the second step. The scope of the research is meaningful in reducing the cost of collecting data by developing predictive-based models that can be used in the stock market and selecting the variables needed for forecasting. The research methodology will use CART (Classification And Regression Tree). In this study, we applied to the stock price prediction in two directions of both prediction and classification. We use the KOSPI 200 data for empirical analysis and present a quantitative forecasting model that considers the direction of future share prices by using financial statement indices, past and present stock prices related to asset value and profitability value. Experiments were conducted in three experiments according to the input parameters of prediction and classification problem. Based on the experimental results, two determinants and two models are suggested for prediction and classification.