Analysis and Prediction of Stock Price Using Hybridization of SARIMA and XGBoost

In the field of stock exchange, to accurately forecast the publicly listed stocks price exchange where many investors use prediction techniques to decide when to invest their money and in which stock they have to invest. Generally, the forecast of publicly listed stock prices is done with the assist...

Full description

Saved in:
Bibliographic Details
Published in2022 International Conference on Communication, Computing and Internet of Things (IC3IoT) pp. 1 - 4
Main Authors Kumar, D.Sathish, C, Thiruvarangan B, C, Sai Nikhil Reddy, A, Vishnu, Devi, A.Sangeerani, Kavitha, D.
Format Conference Proceeding
LanguageEnglish
Published IEEE 10.03.2022
Subjects
ARIMA
Computational modeling
Machine learning
Machine learning algorithms
Predictive models
RMSE
SARIMA
Stock markets
Telecommunication computing
Training
XGBoost
Yahoo finance
Online AccessGet full text

Cover

More Information
Summary:In the field of stock exchange, to accurately forecast the publicly listed stocks price exchange where many investors use prediction techniques to decide when to invest their money and in which stock they have to invest. Generally, the forecast of publicly listed stock prices is done with the assistance of machine learning where it takes into account the present as well as the past price data. It is possible to anticipate stock prices using a variety of algorithms, however standard techniques like data mining statistical or non-deep neural networks are not likely suited to the task because of the volatility and unpredictable behavior of publicly listed stock prices. A stock data set of ten years from Yahoo Finance was used to estimate the values using machine learning that applies a variety of algorithms. Predicting the value of publicly traded stocks using machine learning based on SARIMA and XGBoost is the topic of this study. When predicting the stock market, seasonality plays an important part in the SARIMA model, which is similar to an ARIMA model. Open, adjusted end of the day price, day's peak, day's low, and total volume are all taken into account in XGBoost, an implementation of gradient boosted decision trees optimized for speed and performance. As compared to standard forecasting approaches, the suggested SARIMA XGBoost hybrid model shows Accuracy of 89.48%, Mean Absolute Error (MAE) of 15.612 and Mean Absolute Percentage Error (MAPE) of 10.52%
DOI:10.1109/IC3IOT53935.2022.9767868