A Hybrid Machine Learning Technique for Stork Market Prices Prediction

Stock markets in financial domain play crucial role in economy of nations in the contemporary world. There are many stakeholders who depend on the prediction of stock prices for making buying and selling decisions. In this context, it became important to deal with automatic stock prices prediction. Since manual observation is not possible, machine learning (ML) techniques are widely used for prediction of stock prices movement. Many ML algorithms came into existence. However, their performance largely depends on the quality of training data as they are based on unsupervised learning. In this paper, a hybrid ML framework is proposed to have both feature selections to improve quality of training and the leveraging forecasting performance. The proposed framework supports different techniques such as Linear Regression, XG Boost Regression and Gradient Boost. A prototype application is built using Python data science platform. Experimental results revealed that the three prediction models are providing high level of accuracy. Linear Regression showed 99.989% accuracy, GBoost regression 99.981 and XGBoost Reggressor 0.99969. From the results, it is ascertained that the proposed framework is useful for efficient forecasting of stock prices movements.