Performance Evaluation of Machine Learning Classifiers for Stock Market Prediction in Big Data Environment

Abstract

Implementing machine learning models for the stock's big data emerged as a component of algorithmic trading systems. This paper proposed a hybrid stock prediction model based on the collection of qualitative and quantitative data of particular stocks. In addition to tweets and news data, product reviews of the specific companies traded under National Stock Exchange are considered to analyze their effect on the stock movements. Historical Prices will be integrated with sentiment values generated from tweets, news and product reviews data to construct the amalgam model using Apache Spark and HDFS for storage of large data. The proposed model has been implemented in Google Cloud Platform with different cluster configurations. The paper compares the prediction accuracy based on various types of input data provided to the model using some popular machine learning algorithms.