SELECTION OF IMPORTANCE INDICATORS FOR MACHINE LEARNING MODELS IN FOREX TRADING AREA

Abstract

How to choose the best input variable for use in machine learning is the big question. In real life, the selection of indicators will help improve the results of forex market trend prediction, stock market based on machine learning models is always a topic of great interest to many scientists and investors. In this article, we focus on solving the problem of how to select the best indicators based on Random Uniform Forest. Our method consists of 3 steps: First, We collect data including indices commonly used in the forex sector; second, the data is standardized and labeled; finally, We use Random Uniform Forests to select indicators that are beneficial for prediction. Through the method done, In 17 common indicators in our interested domain, we found out 5 indicators (vol, cci, adx, ar and chv) are most important. We can explain why those indicators is beneficial for machine learning models, improving the model's performance, computation speed and reduced number of data dimensions.