Stock Trend Prediction using Ensemble Learning Techniques in Python
P. Rajesh1, N. Srinivas2, K. Vamshikrishna Reddy3, G. Vamsi Priya4, Vakula Dwija.M5, D. Himaja6
1Dr. P. Rajesh, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
2Dr. N. Srinivasu, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
3K. Vamshi Krishna Reddy, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
4G. Vamsi Priya, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
5Vakula Dwija. M, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
6D. Himaja, Department of Computer Science, Koneru Lakshmaiah Education Foundation, Vaddeswaram (A.P), India.
Manuscript received on 07 March 2019 | Revised Manuscript received on 20 March 2019 | Manuscript published on 30 March 2019 | PP: 150-155 | Volume-8 Issue-5, March 2019 | Retrieval Number: E2922038519/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Stock trends are generated in huge volume and it changes every second. Stock market is a complex and volatile system where people will either gain money or lose their entire life savings. This project is about taking quantifiable data from finance API about the top 500 companies in S&P stock exchange and predicting its future stock trend with ensemble learning. To achieve it we have considered mainly two prediction methods, Heat Map and Ensemble Learning, which based on the percentage change in the stock price data will classify the stock into buy, sell or hold categories. Heat map is generated based on the correlation coefficient of the quantifiable data to further classify the stock as one of the three above mentioned categories. On the other hand, we used the ensemble learning model to classify the stock into a majority vote-based system that considers 3 main classification models. Observations shows that Random Forest, SVM and K-neighbors classifiers show the most prominent results of all other possible combinations. The accuracy of the prediction model is more than 51% whereas in comparison with prediction models with a single classifier labelling with 30% accuracy the model has increased the accuracy by 23%.
Keyword: Stock Trends, Machine Learning, Ensemble Learning, Heat Map, K-Neighbors, Random Forest, SVM.
Scope of the Article: Machine Learning