Ahmed BaniMustafa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Ahmed BaniMustafa has a total of 30 co-authors. Here are the top co-authors by number of shared publications:
Olla Bulkrock
Data Science Department, Princess Sumaya University for Technology, Amman, Jordan
4 shared publicationsNigel Hardy
3 shared publicationsGhassan Samara
2 shared publicationsMohammad Aljaidi
2 shared publicationsSattam Almatarneh
2 shared publications- Abdallah Qusef
Princess Sumaya University for Technology, Amman, Jordan
1 shared publication
Latest Publications
- Open Access
Mode-Dependent Performance of ARIMA, SARIMA, Holt-Winters, and Prophet in Electricity Load Forecasting: A Held-Out Comparison in Jordan
Accurate forecasting of electricity demand is important for generation and distribution planning, especially in Jordan, where distinct seasons, variable weather, a fast-growing economy, a rising population, and shifting consumption habits affect demand. This paper applies four univariate time-series methods (ARIMA, SARIMA, Holt-Winters triple exponential smoothing, and Prophet) to one year of aggregated national daily electricity consumption data collected by the Jordanian Electric Power Company (JEPCO). The forecasting models are trained and then tested on two horizons that span held-out periods of three months and six months. Fitting is performed both with dynamic rolling multi-step forecasts across the entire held-out period and with one-step-ahead out-of-sample forecasts whose parameters are frozen at their training-window values. All models are compared with a weekly seasonal-naive forecast, and pairwise accuracy differences are assessed with Diebold–Mariano tests. The main finding in this work shows that model performance is dependent on evaluation mode. In one-step-ahead mode, Holt-Winters is the most accurate model on both horizons (MAPE of 1.90% and 2.27%; ); weekly SARIMA is statistically indistinguishable from it at three months, whereas ARIMA and Prophet are significantly less accurate. In dynamic multi-step mode, accuracy degrades sharply across all models (MAPE of 7.7–16.8%); no model attains a positive held-out R2; and none significantly outperforms the seasonal-naive benchmark. Exponential smoothing with weekly seasonality is therefore the strongest univariate choice for short-lead operational prediction, whereas reliable medium-term multi-step forecasting requires exogenous information.Open AccessMode-Dependent Performance of ARIMA, SARIMA, Holt-Winters, and Prophet in Electricity Load Forecasting: A Held-Out Comparison in Jordan
Accurate forecasting of electricity demand is important for generation and distribution planning, especially in Jordan, where distinct seasons, variable weather, a fast-growing economy, a rising population, and shifting consumption habits affect demand. This paper applies four univariate time-series methods (ARIMA, SARIMA, Holt-Winters triple exponential smoothing, and Prophet) to one year of aggregated national daily electricity consumption data collected by the Jordanian Electric Power Company (JEPCO). The forecasting models are trained and then tested on two horizons that span held-out periods of three months and six months. Fitting is performed both with dynamic rolling multi-step forecasts across the entire held-out period and with one-step-ahead out-of-sample forecasts whose parameters are frozen at their training-window values. All models are compared with a weekly seasonal-naive forecast, and pairwise accuracy differences are assessed with Diebold–Mariano tests. The main finding in this work shows that model performance is dependent on evaluation mode. In one-step-ahead mode, Holt-Winters is the most accurate model on both horizons (MAPE of 1.90% and 2.27%; ); weekly SARIMA is statistically indistinguishable from it at three months, whereas ARIMA and Prophet are significantly less accurate. In dynamic multi-step mode, accuracy degrades sharply across all models (MAPE of 7.7–16.8%); no model attains a positive held-out R2; and none significantly outperforms the seasonal-naive benchmark. Exponential smoothing with weekly seasonality is therefore the strongest univariate choice for short-lead operational prediction, whereas reliable medium-term multi-step forecasting requires exogenous information. - 17 April 2025
- Published by Springer Nature
- p. 377-388
Predicting Product Rating Based on Customer Reviews and Feedback
- 17 April 2025
- Published by Springer Nature
- p. 377-388
Predicting Product Rating Based on Customer Reviews and Feedback
- Open Access
- 15 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 28280-28293
Reconstruction and Visualization of 5μm Sectional Coronal Views for Macula Vasculature in OptoVue OCTA
A Computerized Medical Image Processing (CMIP) method is proposed to address the current challenges of optical coherence tomography angiography (OCTA): (1) the need for observing the macula vasculature concerning natural curvature of the macula region; (2) the need for generating OCTA frames at successive small depths in all macula layers; and (3) the need for enhancing the visibility of blood vessels, particularly below the outer retina region. The proposed CMIP method involves image preprocessing, reconstruction, and enhancement stages. Twenty subjects were obtained from the OCTA500 dataset, which was obtained from the OptoVue OCTA machine. The 20 subjects comprise the two OCTA fields of view (FOV), right and left eyes (OD and OS), and five common macula disorders. The sequential enface OCTA images at 5μm macula depths were displayed. The presentation of the macula vasculature was enhanced at all depths. The resulting new ophthalmic views enable: (1) avoiding the superimposition of macula vasculature into a projection map; (2) enhancing the OCTA presentation of blood vessels; and (3) inspecting the macula’s 3D oval-shaped. The proposed CMIP method can generate sectional macula coronal views (MCV) for every 5μm depth, clarifying the macula’s curvature in a way that has not been presented in OCTA machines. Therefore, “tracking” the 3D propagation of the blood vessel network at all macula depths becomes possible. Furthermore, the blood vessels’ display at all macula depths, including the deep choroid layers, is enhanced. The method yields futuristic ophthalmic advantages that would allow the physician to precisely inspect the 3D localization and diffusion of the macula disorders. The method is invariant to the OCTA’s FOVs, macula disorder, and OD or OS eye.
- 15 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 28280-28293
Open AccessReconstruction and Visualization of 5μm Sectional Coronal Views for Macula Vasculature in OptoVue OCTA
A Computerized Medical Image Processing (CMIP) method is proposed to address the current challenges of optical coherence tomography angiography (OCTA): (1) the need for observing the macula vasculature concerning natural curvature of the macula region; (2) the need for generating OCTA frames at successive small depths in all macula layers; and (3) the need for enhancing the visibility of blood vessels, particularly below the outer retina region. The proposed CMIP method involves image preprocessing, reconstruction, and enhancement stages. Twenty subjects were obtained from the OCTA500 dataset, which was obtained from the OptoVue OCTA machine. The 20 subjects comprise the two OCTA fields of view (FOV), right and left eyes (OD and OS), and five common macula disorders. The sequential enface OCTA images at 5μm macula depths were displayed. The presentation of the macula vasculature was enhanced at all depths. The resulting new ophthalmic views enable: (1) avoiding the superimposition of macula vasculature into a projection map; (2) enhancing the OCTA presentation of blood vessels; and (3) inspecting the macula’s 3D oval-shaped. The proposed CMIP method can generate sectional macula coronal views (MCV) for every 5μm depth, clarifying the macula’s curvature in a way that has not been presented in OCTA machines. Therefore, “tracking” the 3D propagation of the blood vessel network at all macula depths becomes possible. Furthermore, the blood vessels’ display at all macula depths, including the deep choroid layers, is enhanced. The method yields futuristic ophthalmic advantages that would allow the physician to precisely inspect the 3D localization and diffusion of the macula disorders. The method is invariant to the OCTA’s FOVs, macula disorder, and OD or OS eye.
- Open Access
- 9 February 2023
- Published by Agora University of Oradea in International Journal of Computers Communications & Control
- Vol. 18 (1) , 17
Deep Learning for Assessing Severity of Cracks in Concrete Structures
Most concrete structures suffer from degradation, where cracks are the most obvious visual sign. Concrete structures must be continuously monitored and assessed to avoid further deterioration, which may lead to a partial or total collapse. This is particularly important when constructing large structures such as towers, bridges, tunnels, and dams. This work aims to demonstrate and evaluate several deep learning approaches that can be used for monitoring and assessing the level of concrete degradation based on the cracks’ visual signs, which can then be embedded in Health Monitoring Systems (SHM). The experimental work in this study involves creating three models: Two were built using ResNet-50 and Xception transfer learning networks. In contrast, the third was built using a customized Sequential Convolutional Neural Network (SCNN) architecture. The dataset comprises 2,000 image samples sampled from a larger dataset that contains 56,000 images and which belong to four severity classes: minor, moderate, and severe, in addition to a normal class for no crack signs. The SCNN model achieved an accuracy of 90.2%, while the Xception and ResNet-50 models scored an accuracy of 86.3% and 70%, respectively.
- 9 February 2023
- Published by Agora University of Oradea in International Journal of Computers Communications & Control
- Vol. 18 (1) , 17
Open AccessDeep Learning for Assessing Severity of Cracks in Concrete Structures
Most concrete structures suffer from degradation, where cracks are the most obvious visual sign. Concrete structures must be continuously monitored and assessed to avoid further deterioration, which may lead to a partial or total collapse. This is particularly important when constructing large structures such as towers, bridges, tunnels, and dams. This work aims to demonstrate and evaluate several deep learning approaches that can be used for monitoring and assessing the level of concrete degradation based on the cracks’ visual signs, which can then be embedded in Health Monitoring Systems (SHM). The experimental work in this study involves creating three models: Two were built using ResNet-50 and Xception transfer learning networks. In contrast, the third was built using a customized Sequential Convolutional Neural Network (SCNN) architecture. The dataset comprises 2,000 image samples sampled from a larger dataset that contains 56,000 images and which belong to four severity classes: minor, moderate, and severe, in addition to a normal class for no crack signs. The SCNN model achieved an accuracy of 90.2%, while the Xception and ResNet-50 models scored an accuracy of 86.3% and 70%, respectively.
- 4 February 2023
- Published by IJETAE Publication House in International Journal of Emerging Technology and Advanced Engineering
- Vol. 13 (2) , 40-46
Data Collection in WSNs using a Probability-Based Rendezvous Points Selection Algorithm
Wireless Sensor Networks are becoming more prevalent in various industries, including military operations and distant environmental monitoring. This is important because sensors are getting smarter, smaller, and less expensive. The energy hole problem in the WSN has been a major focus of recent research. The mobile sink is an efficient solution for the energy hole problem in a wireless sensor network. A mobile sink gets data from sensors by moving around the network often to avoid problems with hotspots or energy holes. It gets data from network nodes by traveling regularly and visiting a group of nodes known as rendezvous points (RPs). This research will present a probability-based RP selection (PRPS) technique for data collection in wireless sensor networks. To begin, a directed spanning tree is used to construct a tree that eliminates duplication in the data forwarding path. The proposed method is employed to compute the likelihood of RPs. Finally, using the shortest path technique, a mobile sink is constructed between these locations. The path provided is the best path that connects all of the RPs. The proposed approach improves the previous solutions by choosing the nodes with the most data packets as RPs. As a result, it extends network lifetime by lowering energy consumption and addressing the energy hole problem.
- 4 February 2023
- Published by IJETAE Publication House in International Journal of Emerging Technology and Advanced Engineering
- Vol. 13 (2) , 40-46
Data Collection in WSNs using a Probability-Based Rendezvous Points Selection Algorithm
Wireless Sensor Networks are becoming more prevalent in various industries, including military operations and distant environmental monitoring. This is important because sensors are getting smarter, smaller, and less expensive. The energy hole problem in the WSN has been a major focus of recent research. The mobile sink is an efficient solution for the energy hole problem in a wireless sensor network. A mobile sink gets data from sensors by moving around the network often to avoid problems with hotspots or energy holes. It gets data from network nodes by traveling regularly and visiting a group of nodes known as rendezvous points (RPs). This research will present a probability-based RP selection (PRPS) technique for data collection in wireless sensor networks. To begin, a directed spanning tree is used to construct a tree that eliminates duplication in the data forwarding path. The proposed method is employed to compute the likelihood of RPs. Finally, using the shortest path technique, a mobile sink is constructed between these locations. The path provided is the best path that connects all of the RPs. The proposed approach improves the previous solutions by choosing the nodes with the most data packets as RPs. As a result, it extends network lifetime by lowering energy consumption and addressing the energy hole problem.
- Open Access
- 31 January 2023
- Published by The Science and Information Organization in International Journal of Advanced Computer Science and Applications
- Vol. 14 (1)
A Machine Learning Hybrid Approach for Diagnosing Plants Bacterial and Fungal Diseases
Bacterial and Fungal diseases may affect the yield of stone fruit and cause damage to the Chlorophyll synthesis process, which is crucial for tree growth and fruiting. However, due to their similar visual shot-hole symptoms, novice agriculturalists and ordinary farmers usually cannot identify and differentiate these two diseases. This work investigates and evaluates the use of machine learning for diagnosing these two diseases. It aims at paving the way toward creating a generic deep learning-based model that can be embedded in a mobile phone application or in a web service to provide a fast, reliable, and cheap diagnosis for plant diseases which help reduce the excessive, unnecessary, or improper use of pesticides, which can harm public health and the environment. The dataset consists of hundreds of samples collected from stone fruit farms in the north of Jordan under normal field conditions. The image features were extracted using a CNN algorithm that was pre-trained with millions of images, and the diseases were identified using three machine learning classification algorithms: 1) K-nearest neighbour (KNN); 2) Stochastic Gradient Descent (SGD); and 3) Random Forests (RF). The resulting models were evaluated using 10-fold cross-validation, with CNN-KNN achieving the best AUC performance with a score of 98.5%. On the other hand, the CNN-SGD model performed best in Classification Accuracy (CA) with a score of 93.7%. The results shown in the Confusion Matrix, ROC, Lift, and Calibration curves also confirmed the validity and robustness of the constructed models.
- 31 January 2023
- Published by The Science and Information Organization in International Journal of Advanced Computer Science and Applications
- Vol. 14 (1)
Open AccessA Machine Learning Hybrid Approach for Diagnosing Plants Bacterial and Fungal Diseases
Bacterial and Fungal diseases may affect the yield of stone fruit and cause damage to the Chlorophyll synthesis process, which is crucial for tree growth and fruiting. However, due to their similar visual shot-hole symptoms, novice agriculturalists and ordinary farmers usually cannot identify and differentiate these two diseases. This work investigates and evaluates the use of machine learning for diagnosing these two diseases. It aims at paving the way toward creating a generic deep learning-based model that can be embedded in a mobile phone application or in a web service to provide a fast, reliable, and cheap diagnosis for plant diseases which help reduce the excessive, unnecessary, or improper use of pesticides, which can harm public health and the environment. The dataset consists of hundreds of samples collected from stone fruit farms in the north of Jordan under normal field conditions. The image features were extracted using a CNN algorithm that was pre-trained with millions of images, and the diseases were identified using three machine learning classification algorithms: 1) K-nearest neighbour (KNN); 2) Stochastic Gradient Descent (SGD); and 3) Random Forests (RF). The resulting models were evaluated using 10-fold cross-validation, with CNN-KNN achieving the best AUC performance with a score of 98.5%. On the other hand, the CNN-SGD model performed best in Classification Accuracy (CA) with a score of 93.7%. The results shown in the Confusion Matrix, ROC, Lift, and Calibration curves also confirmed the validity and robustness of the constructed models.
- Open Access
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 3306 (1) , 040001
Random forests for predicting software effort estimation based on use-case points analysis
Software estimation is vital for the success of software engineering projects. However, predicting software effort is difficult due to software solutions’ compl
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 3306 (1) , 040001
Open AccessRandom forests for predicting software effort estimation based on use-case points analysis
Software estimation is vital for the success of software engineering projects. However, predicting software effort is difficult due to software solutions’ compl
- Open Access
- 31 December 2022
- Published by The Science and Information Organization in International Journal of Advanced Computer Science and Applications
- Vol. 13 (12)
Machine Learning for Securing Traffic in Computer Networks
Computer network attacks are among the most significant and common threats against computer-wired and wireless communications. Intrusion detection technology is used to secure computer networks by monitoring network traffic and identifying attacks. In this paper, we investigate and evaluate the application of four machine learning classification algorithms for identifying attacks that target computer networks: DDoS, Brute Force Web, and SQL Injection attacks, in addition to Benign Traffic. A public dataset of 80 features was used to build four machine learning models using Random Forest, Logistic Regression, CN2, and Neural Networks. The constructed models were evaluated based on 10-fold cross-validation using Classification Accuracy (CA), Area under the Curve (AUC), F1, Recall, Specificity, and Sensitivity metrics in addition to Confusion Matrix, Calibration, Lift, and ROC plots. The Random Forest model achieved 98% in the CA score and 99% in the AUC score, while the Logistic regression achieved 90% in the CA score and 98% in the AUC score.
- 31 December 2022
- Published by The Science and Information Organization in International Journal of Advanced Computer Science and Applications
- Vol. 13 (12)
Open AccessMachine Learning for Securing Traffic in Computer Networks
Computer network attacks are among the most significant and common threats against computer-wired and wireless communications. Intrusion detection technology is used to secure computer networks by monitoring network traffic and identifying attacks. In this paper, we investigate and evaluate the application of four machine learning classification algorithms for identifying attacks that target computer networks: DDoS, Brute Force Web, and SQL Injection attacks, in addition to Benign Traffic. A public dataset of 80 features was used to build four machine learning models using Random Forest, Logistic Regression, CN2, and Neural Networks. The constructed models were evaluated based on 10-fold cross-validation using Classification Accuracy (CA), Area under the Curve (AUC), F1, Recall, Specificity, and Sensitivity metrics in addition to Confusion Matrix, Calibration, Lift, and ROC plots. The Random Forest model achieved 98% in the CA score and 99% in the AUC score, while the Logistic regression achieved 90% in the CA score and 98% in the AUC score.
- 7 December 2022
A Data Science Approach for Predicting Crowdfunding Success
- 6 December 2022
Trends of Electricity Consumption in Jordan