|
|
 |
| |
|
|
|
Analysis of the Stock Market Using the Integration of Statistical and Machine Learning Models |
|
|
|
PP: 115-122 |
|
|
|
doi:10.18576/jsap/140109
|
|
|
|
Author(s) |
|
|
|
M. Aripov,
Azhari A. Alhag,
Siham A. Shaddad,
Nadia B. Mohammed Ali,
Hiba A. A. A. Hussin,
|
|
|
|
Abstract |
|
|
| Time series prediction is a critical task in various fields, including finance, economics, and field of finance. In this study, we assess the forecasting performance of three distinct models-Artificial Neural Networks (ANN), Autoregressive Integrated Moving Average (ARIMA), and a Hybrid model-using a dataset of Saudi Basic Industries Corporation (SABIC) stock prices, covering the period from January 1, 2016, to August 10, 2024. The models are evaluated based on three widely recognized error metrics: Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), and Symmetric Mean Absolute Percentage Error (sMAPE). The Hybrid model, which integrates multiple forecasting approaches, consistently outperforms both the ANN and ARIMA models across all three metrics. The results reveal that the Hybrid model provides the most accurate and stable predictions, with significantly lower error values, including a notably lower coefficient of variation (CoefVar) compared to the other models. The ANN model, while effective, exhibits slightly higher variability and error rates, while ARIMA struggles to capture extreme values in the data. Boxplots of actual and predicted values demonstrate that all models successfully capture the general trends in the data without producing substantial outliers. Based on these findings, the Hybrid model is recommended for stock price forecasting, particularly when prediction accuracy, stability, and minimal variability are prioritized.
|
|
|
|
|
|
|
|
