Stock Price’s Prediction with Decision Tree
Establishment of one process and some ameliorations of decision tree’s algorithm in order to predict the second day’s price change. The experiment builds a J48 tree, which is comfortable with continuous attributes, based on 10 years historical stock prices. After careful selection and preprocessing of financial data, high prediction accuracy is obtained. An introduction of dynamic-constructed tree reduces tree’s cost, and increases prediction’s quality on accuracy as well as average error distance.
T. Li and G. S. Liu, "Stock Price’s Prediction with Decision Tree", Applied Mechanics and Materials, Vols. 48-49, pp. 1116-1121, 2011