Cafer Budak
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Cafer Budak has a total of 28 co-authors. Here are the top co-authors by number of shared publications:
- Veysel Gider
Distance Education Application and Research Center, Rectorate, Batman University
4 shared publications Vasfiye Mençik
Department of Electric-Electronic Engineering, Dicle University, Diyarbakır, Turkey
3 shared publicationsVeysel Gider
Department of Electric-Electronic Engineering, Dicle University, Diyarbakır, Turkey; Distance Education Application and Research Center, Rectorate, Batman University, Batman, Turkey
2 shared publications- Abdurrahman Yilmaz
Department of Control and Automation Engineering, Istanbul Technical University, Istanbul 34485, Türkiye
1 shared publication - Ali Cinar
Kastamonu University
1 shared publication - Ali Mardani
Department of Civil Engineering, Faculty of Engineering, Bursa Uludag University, Bursa 16059, Türkiye
1 shared publication
Latest Publications
- Open Access
Predicting the Workability of PCE-Modified Cement–Fly Ash Pastes with Explainable Machine Learning
Reliable assessment of polycarboxylate ether (PCE)–binder combinations requires predictive models whose interpolation performance is distinguished from their performance when an entire formulation is absent from training. This study examined 616 cement–fly ash pastes prepared using twenty-two in-house PCE formulations, four fly ash replacement levels (0, 15, 30, and 45 wt %), seven PCE dosages (0.50–2.00 wt % of binder), and a fixed water-to-binder ratio of 0.35. Marsh funnel flow time was measured for all mixtures, whereas mini-slump measurements were available for 252 mixtures comprising nine PCE formulations. Ten regression algorithms were evaluated using 3 × 5-fold repeated cross-validation and nine non-algebraically redundant input variables. XGBoost achieved R2 = 0.946 ± 0.027 and RMSE = 8.40 s for flow time, and R2 = 0.778 ± 0.061 and RMSE = 0.471 cm for mini-slump. These random-resampling results describe interpolation among formulations represented in the training folds. When each PCE chemistry was withheld in turn, the mean R2 was 0.865 ± 0.198 for flow time and 0.034 ± 0.530 for mini-slump. Performance also decreased when the boundary fly ash levels were withheld. SHAP and permutation analyses identified PCE dose and fly ash replacement as the strongest predictors, while the formulation-level descriptors made smaller contributions. These findings represent associations learned from the present dataset and do not constitute direct evidence of adsorption or dispersion mechanisms. At a nominal 90% level, split-conformal intervals achieved empirical coverages of 85.5% for flow time—moderately below the nominal level, within finite-sample binomial fluctuation—and 90.2% for mini-slump on a random test partition; across fifty repeated train–calibration–test splits, the mean coverages were 89.4 ± 3.6% and 90.1 ± 5.8%. The flow-time model is suitable for preliminary screening within the investigated factor ranges, whereas the mini-slump model should be restricted to interpolation among the sampled formulations.
Open AccessPredicting the Workability of PCE-Modified Cement–Fly Ash Pastes with Explainable Machine Learning
Reliable assessment of polycarboxylate ether (PCE)–binder combinations requires predictive models whose interpolation performance is distinguished from their performance when an entire formulation is absent from training. This study examined 616 cement–fly ash pastes prepared using twenty-two in-house PCE formulations, four fly ash replacement levels (0, 15, 30, and 45 wt %), seven PCE dosages (0.50–2.00 wt % of binder), and a fixed water-to-binder ratio of 0.35. Marsh funnel flow time was measured for all mixtures, whereas mini-slump measurements were available for 252 mixtures comprising nine PCE formulations. Ten regression algorithms were evaluated using 3 × 5-fold repeated cross-validation and nine non-algebraically redundant input variables. XGBoost achieved R2 = 0.946 ± 0.027 and RMSE = 8.40 s for flow time, and R2 = 0.778 ± 0.061 and RMSE = 0.471 cm for mini-slump. These random-resampling results describe interpolation among formulations represented in the training folds. When each PCE chemistry was withheld in turn, the mean R2 was 0.865 ± 0.198 for flow time and 0.034 ± 0.530 for mini-slump. Performance also decreased when the boundary fly ash levels were withheld. SHAP and permutation analyses identified PCE dose and fly ash replacement as the strongest predictors, while the formulation-level descriptors made smaller contributions. These findings represent associations learned from the present dataset and do not constitute direct evidence of adsorption or dispersion mechanisms. At a nominal 90% level, split-conformal intervals achieved empirical coverages of 85.5% for flow time—moderately below the nominal level, within finite-sample binomial fluctuation—and 90.2% for mini-slump on a random test partition; across fifty repeated train–calibration–test splits, the mean coverages were 89.4 ± 3.6% and 90.1 ± 5.8%. The flow-time model is suitable for preliminary screening within the investigated factor ranges, whereas the mini-slump model should be restricted to interpolation among the sampled formulations.
- 1 June 2026
- Published by Gazi University Journal of Science in Gazi University Journal of Science
- Vol. 39 (2) , 939-955
Environmental Risk Assessment and Resource Recovery Potential of Asphaltite Ash in Şırnak (Türkiye): A Geochemical and Machine Learning Approach
In this study, asphaltite samples obtained from five different sites (Harbul, Milli, Avgamasya, Seridahlı and Karatepe) in Şırnak province were ashes and analyzed by ICP-OES method after microwave-assisted acid solubilization. According to the results of the analysis, the abundance order of the major elements was determined as Fe (52.400-61.200 mg/kg) > Zn (10.860-39.805 mg/kg) > Ni (2.532-4.755 mg/kg) > Mo (2.246-3.186 mg/kg), and it was confirmed that the relative standard deviation (RSD) values in repeated measurements were below 5%. Analyses revealed that especially Cd, Mo, Se and Zn elements were enriched at high levels in all samples. Assessments using the Metal Pollution Index (MPI), Geoaccumulation Index (Igeo) and Potential Ecological Risk Index (RI) showed that all sites were in the very high ecological risk class (RI>600). It has been determined that the Cd-induced risk is dominant, especially in the Harbul area. In contrast, the high concentrations determined in terms of Mo and Zn at the Seridahlı site indicate the economic recovery potential of this ash; according to calculations, the metal recovery value of Seridahlı ash is approximately 323 USD/ton. Geochemical network analysis revealed strong genetic relationships between Mo-Zn-Cu-Se elements, while regression-based machine learning models confirmed that Cd acts independently of other elements and its critical role in terms of environmental risk. The findings show that Şırnak asphaltite ashes pose a serious environmental risk on the one hand, and on the other hand have a significant secondary raw material potential that can be evaluated with appropriate management processes.
- 1 June 2026
- Published by Gazi University Journal of Science in Gazi University Journal of Science
- Vol. 39 (2) , 939-955
Environmental Risk Assessment and Resource Recovery Potential of Asphaltite Ash in Şırnak (Türkiye): A Geochemical and Machine Learning Approach
In this study, asphaltite samples obtained from five different sites (Harbul, Milli, Avgamasya, Seridahlı and Karatepe) in Şırnak province were ashes and analyzed by ICP-OES method after microwave-assisted acid solubilization. According to the results of the analysis, the abundance order of the major elements was determined as Fe (52.400-61.200 mg/kg) > Zn (10.860-39.805 mg/kg) > Ni (2.532-4.755 mg/kg) > Mo (2.246-3.186 mg/kg), and it was confirmed that the relative standard deviation (RSD) values in repeated measurements were below 5%. Analyses revealed that especially Cd, Mo, Se and Zn elements were enriched at high levels in all samples. Assessments using the Metal Pollution Index (MPI), Geoaccumulation Index (Igeo) and Potential Ecological Risk Index (RI) showed that all sites were in the very high ecological risk class (RI>600). It has been determined that the Cd-induced risk is dominant, especially in the Harbul area. In contrast, the high concentrations determined in terms of Mo and Zn at the Seridahlı site indicate the economic recovery potential of this ash; according to calculations, the metal recovery value of Seridahlı ash is approximately 323 USD/ton. Geochemical network analysis revealed strong genetic relationships between Mo-Zn-Cu-Se elements, while regression-based machine learning models confirmed that Cd acts independently of other elements and its critical role in terms of environmental risk. The findings show that Şırnak asphaltite ashes pose a serious environmental risk on the one hand, and on the other hand have a significant secondary raw material potential that can be evaluated with appropriate management processes.
- Open Access
- 24 March 2026
- Published by Ankara University Faculty of Agriculture in Tarım Bilimleri Dergisi
- Vol. 32 (2) , 439-454
Adaptive Capacity of Machine Learning Approaches for Leaf Area Estimation in Faba Bean Under Diverse Agro-environmental Conditions
In controlled, shaded agricultural environments, conventional remote sensing technologies are often ineffective due to their reliance on direct sunlight and sensitivity to variations in reflectance and absorption. These limitations reduce the effectiveness and precision of image-based leaf area estimation methods, especially in settings such as polytunnel farming, greenhouses, and areas covered by shade netting. Machine learning methods, on the other hand, are used to obtain reliable estimates from complex datasets in various fields, including biology, economics, and engineering. They provide significant practical advantages in applications such as production forecasting and environmental impact analysis in agriculture. This study investigates the use of machine learning models (ElasticNet, Support Vector Regression, Random Forest, Gradient Boosting, and XGBoost) to estimate the leaf area of faba bean cultivars grown under different shading levels. The study used non-destructive morphological traits, such as plant height, leaf number, and leaf density index, as predictive variables. The dataset, incorporated diverse environmental and biological factors, including varying shading intensities, plant densities, and growth stages across different years. The Support Vector Regression (SVR) model achieved the highest predictive accuracy (R² = 0.92, RMSE = 0.07), significantly outperforming the linear ElasticNet model (R² = 0.76). The proposed model provides a low cost, efficient, and non-destructive approach to leaf area estimation in shaded environments where conventional remote sensing is limited.
- 24 March 2026
- Published by Ankara University Faculty of Agriculture in Tarım Bilimleri Dergisi
- Vol. 32 (2) , 439-454
Open AccessAdaptive Capacity of Machine Learning Approaches for Leaf Area Estimation in Faba Bean Under Diverse Agro-environmental Conditions
In controlled, shaded agricultural environments, conventional remote sensing technologies are often ineffective due to their reliance on direct sunlight and sensitivity to variations in reflectance and absorption. These limitations reduce the effectiveness and precision of image-based leaf area estimation methods, especially in settings such as polytunnel farming, greenhouses, and areas covered by shade netting. Machine learning methods, on the other hand, are used to obtain reliable estimates from complex datasets in various fields, including biology, economics, and engineering. They provide significant practical advantages in applications such as production forecasting and environmental impact analysis in agriculture. This study investigates the use of machine learning models (ElasticNet, Support Vector Regression, Random Forest, Gradient Boosting, and XGBoost) to estimate the leaf area of faba bean cultivars grown under different shading levels. The study used non-destructive morphological traits, such as plant height, leaf number, and leaf density index, as predictive variables. The dataset, incorporated diverse environmental and biological factors, including varying shading intensities, plant densities, and growth stages across different years. The Support Vector Regression (SVR) model achieved the highest predictive accuracy (R² = 0.92, RMSE = 0.07), significantly outperforming the linear ElasticNet model (R² = 0.76). The proposed model provides a low cost, efficient, and non-destructive approach to leaf area estimation in shaded environments where conventional remote sensing is limited.
- Open Access
- 18 February 2025
- Published by Fırat University, Faculty of Engineering in Firat University Journal of Experimental and Computational Engineering
- Vol. 4 (1) , 1-11
Recommending Tolerance Value for SpO2 Devices with Linear Regression Based on Measuring Tape
Monitoring blood oxygen levels is vital for tracking various respiratory diseases like bronchitis, pneumonia, COPD, and critical care patients, including those with COVID-19. SpO2 devices, calculating oxygen percentage via finger or earlobe tissue, play a crucial role. Field studies have revealed concerns regarding the accuracy of SpO2 measurements due to high deviations, particularly in patients with low oxygen saturation, prompting the initiation of this study to ensure accurate interpretation of the device's measurements. Using a linear regression algorithm, SpO2 values from different bands were classified for quality. Tolerance values and deviation thresholds for each band were recommended. Additionally, linear regression aimed to save time by making result estimations with less data, facilitating more device monitoring and frequent testing. Results closely matched actual values, suggesting contributions to more frequent application and rapid interpretation. Following the European standard 80601-2-61, measurements were taken from three pulse oximeter brands (Contec MS100 model simulator) in three bands: 70-79%, 80-89%, and 90-100% SpO2. For each measurement, an Arms curve graph was generated using linear regression, and the mean square error (MSE) and Arms values were calculated to evaluate devices. In conclusion, deviation rates increase at low oxygen saturation levels, and recommended % SpO2 deviation values were proposed for each band and device quality.
- 18 February 2025
- Published by Fırat University, Faculty of Engineering in Firat University Journal of Experimental and Computational Engineering
- Vol. 4 (1) , 1-11
Open AccessRecommending Tolerance Value for SpO2 Devices with Linear Regression Based on Measuring Tape
Monitoring blood oxygen levels is vital for tracking various respiratory diseases like bronchitis, pneumonia, COPD, and critical care patients, including those with COVID-19. SpO2 devices, calculating oxygen percentage via finger or earlobe tissue, play a crucial role. Field studies have revealed concerns regarding the accuracy of SpO2 measurements due to high deviations, particularly in patients with low oxygen saturation, prompting the initiation of this study to ensure accurate interpretation of the device's measurements. Using a linear regression algorithm, SpO2 values from different bands were classified for quality. Tolerance values and deviation thresholds for each band were recommended. Additionally, linear regression aimed to save time by making result estimations with less data, facilitating more device monitoring and frequent testing. Results closely matched actual values, suggesting contributions to more frequent application and rapid interpretation. Following the European standard 80601-2-61, measurements were taken from three pulse oximeter brands (Contec MS100 model simulator) in three bands: 70-79%, 80-89%, and 90-100% SpO2. For each measurement, an Arms curve graph was generated using linear regression, and the mean square error (MSE) and Arms values were calculated to evaluate devices. In conclusion, deviation rates increase at low oxygen saturation levels, and recommended % SpO2 deviation values were proposed for each band and device quality.
- Open Access
- 25 March 2024
- Published by Gazi Universitesi Fen Bilimleri Dergisi Part C: Tasarim ve Teknoloji in Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji
- Vol. 12 (1) , 164-175
Drug Solubility Prediction: A Comparative Analysis of GNN, MLP, and Traditional Machine Learning Algorithms
The effective development and design of pharmaceuticals hold fundamental importance in the fields of medicine and the pharmaceutical industry. In this process, the accurate prediction of drug molecule solubility is a critical factor influencing the bioavailability, pharmacokinetics, and toxicity of drugs. Traditionally, mathematical equations based on chemical and physical properties have been used for drug solubility prediction. However, in recent years, with the advancement of artificial intelligence and machine learning techniques, new approaches have been developed in this field. This study evaluated different modeling approaches consisting of Graph Neural Networks (GNN), Multilayer Perceptron (MLP), and traditional Machine Learning (ML) algorithms. The Random Forest (RF) model stands out as the optimal performer, manifesting superior efficacy through the attainment of minimal error rates. It attains a Root Mean Square Error (RMSE) value of 1.2145, a Mean Absolute Error (MAE) value of 0.9221, and an R-squared (R2) value of 0.6575. In contrast, GNN model displays comparatively suboptimal performance, as evidenced by an RMSE value of 1.8389, an MAE value of 1.4684, and an R2 value of 0.2147. These values suggest that the predictions of this model contain higher errors compared to other models, and its explanatory power is lower. These findings highlight the performance differences among different modeling approaches in drug solubility prediction. The RF model is shown to be more effective than other methods, while the GNN model performs less effectively. This information provides valuable insights into which model should be preferred in pharmaceutical design and development processes.
- 25 March 2024
- Published by Gazi Universitesi Fen Bilimleri Dergisi Part C: Tasarim ve Teknoloji in Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım ve Teknoloji
- Vol. 12 (1) , 164-175
Open AccessDrug Solubility Prediction: A Comparative Analysis of GNN, MLP, and Traditional Machine Learning Algorithms
The effective development and design of pharmaceuticals hold fundamental importance in the fields of medicine and the pharmaceutical industry. In this process, the accurate prediction of drug molecule solubility is a critical factor influencing the bioavailability, pharmacokinetics, and toxicity of drugs. Traditionally, mathematical equations based on chemical and physical properties have been used for drug solubility prediction. However, in recent years, with the advancement of artificial intelligence and machine learning techniques, new approaches have been developed in this field. This study evaluated different modeling approaches consisting of Graph Neural Networks (GNN), Multilayer Perceptron (MLP), and traditional Machine Learning (ML) algorithms. The Random Forest (RF) model stands out as the optimal performer, manifesting superior efficacy through the attainment of minimal error rates. It attains a Root Mean Square Error (RMSE) value of 1.2145, a Mean Absolute Error (MAE) value of 0.9221, and an R-squared (R2) value of 0.6575. In contrast, GNN model displays comparatively suboptimal performance, as evidenced by an RMSE value of 1.8389, an MAE value of 1.4684, and an R2 value of 0.2147. These values suggest that the predictions of this model contain higher errors compared to other models, and its explanatory power is lower. These findings highlight the performance differences among different modeling approaches in drug solubility prediction. The RF model is shown to be more effective than other methods, while the GNN model performs less effectively. This information provides valuable insights into which model should be preferred in pharmaceutical design and development processes.
- 1 February 2024
- Published by Springer Nature in Wireless Personal Communications
- Vol. 134 (3) , 1551-1569
Classification of the Ionospheric Disturbances Caused by Geomagnetic and Seismic Activity with K-Nearest Neighbors Algorithm
- 1 February 2024
- Published by Springer Nature in Wireless Personal Communications
- Vol. 134 (3) , 1551-1569
Classification of the Ionospheric Disturbances Caused by Geomagnetic and Seismic Activity with K-Nearest Neighbors Algorithm
- 24 July 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 35 (28) , 20717-20734
Online diagnosis of COVID-19 from chest radiography images by using deep learning algorithms
- 24 July 2023
- Published by Springer Nature in Neural Computing and Applications
- Vol. 35 (28) , 20717-20734
Online diagnosis of COVID-19 from chest radiography images by using deep learning algorithms
- 26 June 2023
- Published by Springer Nature in Earth Science Informatics
- Vol. 16 (3) , 2323-2337
LSTM based forecasting of the next day’s values of ionospheric total electron content (TEC) as an earthquake precursor signal
- 26 June 2023
- Published by Springer Nature in Earth Science Informatics
- Vol. 16 (3) , 2323-2337
LSTM based forecasting of the next day’s values of ionospheric total electron content (TEC) as an earthquake precursor signal
- 30 April 2023
- Published by International Information and Engineering Technology Association in Traitement du Signal
- Vol. 40 (2) , 523-531
Biomedical Image Segmentation with Modified U-Net
- 30 April 2023
- Published by International Information and Engineering Technology Association in Traitement du Signal
- Vol. 40 (2) , 523-531
Biomedical Image Segmentation with Modified U-Net