Financial forecasting using random subspace classifier

Random Subspace Classifier is used for prediction of a stock price return. While obtaining interesting results with a basic model it's possible to construct more competitive network by using several approaches for improving the prediction accuracy and performance characteristics. The following methods are considered in this work: normalizing input data, generating a sensitive classifier structure and variance structure selection. The best average success rate achieved in the prediction of the stock price change direction is 58.1%.