Feature selection methods for financial engineering

Abstract

The analysis about the financial market is always drawing the attention of both the investors and researchers. Theories and methodologies are invented to pattern the stock market finely and easily. The trend of stock market is very complex and is influenced by various factors. Therefore to find out the most significant factors to the stock market is very necessary. Feature Selection is such an algorithm that can remove the redundant and irrelevant factors, and then figure out the most significant subset of factors to build the analysis model.

This project analyzes about a series of technical indicators, which are the results of technical analysis about the stock market. Among them, some may be very significant to the stock price, and others may not have so much influence. These indicators are confined and taken as the input feature set, and then Feature Selection algorithm is used to conduct the selection, and generate a feature subset which contains all the significant features. There are three kinds of Feature Selection algorithms studied in this project. A recommendation based on this research will be given in the end of this report.