p-Index From 2021 - 2026
9.962
P-Index
Claim Missing Document
Check
Articles

Machine Learning Based Cervical Cancer Risk Prediction with SHAP-Driven Feature Interpretation Fachrizal Ardiansyah; Raka Deny Abdi Putra; Budiman
Bulletin of Intelligent Machines and Algorithms Vol. 1 No. 3 (2026): BIMA March 2026 Issue
Publisher : Maheswari Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.65780/bima.v1i3.16

Abstract

Cervical cancer remains a critical public health problem, particularly in developing countries where early detection is often limited. This study presents a machine learning–based approach for cervical cancer risk prediction that emphasizes both predictive accuracy and interpretability. Several supervised algorithms, namely K-Nearest Neighbors, Random Forest, XGBoost, and CatBoost, were evaluated using the Cervical Cancer (Risk Factors) dataset from the UCI Machine Learning Repository following comprehensive data preprocessing and systematic hyperparameter optimization. The experimental results show that CatBoost achieved the best overall performance, with an optimized accuracy of 97.01% and improved sensitivity in detecting high-risk cases, supported by stable k-fold cross-validation results. To enhance clinical transparency, explainable artificial intelligence was incorporated via SHAP, revealing that key predictors such as the Schiller test, age, and reproductive factors played dominant roles in the model’s decisions. These findings demonstrate that the proposed framework is not only accurate and stable but also interpretable and clinically relevant, making it well-suited to support early detection and decision-making in cervical cancer screening, especially in resource-limited healthcare settings.
Comparative Analysis of Machine Learning Regression Models for Paddy Yield Prediction Chery Cardinawati Sitohang; Fitri Kinkin; Budiman
Bulletin of Intelligent Machines and Algorithms Vol. 1 No. 3 (2026): BIMA March 2026 Issue
Publisher : Maheswari Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.65780/bima.v1i3.17

Abstract

Accurate paddy yield prediction is essential to support food security, agricultural planning, and data-driven decision-making. The increasing availability of agricultural data has encouraged the adoption of machine learning approaches to overcome the limitations of conventional yield estimation methods. This study presents a comparative analysis of five regression-based machine learning algorithms—Linear Regression, K-Nearest Neighbors Regressor, Decision Tree Regressor, Random Forest Regressor, and Support Vector Regression—for paddy yield prediction. The experiments were conducted using the Paddy dataset from the UCI Machine Learning Repository, which consists of 2,789 samples and 45 variables (44 input features and 1 target variable). The dataset was preprocessed through data cleaning, feature standardization, and an 80:20 train–test split. Model performance was evaluated using Mean Absolute Error, Mean Squared Error, Root Mean Squared Error, and the coefficient of determination (R²). Experimental results show that Linear Regression achieved the best overall performance with an R² value of 0.9896 and an RMSE of 942.09, indicating strong predictive accuracy and stability. Despite its simplicity, Linear Regression outperformed more complex models, suggesting that the underlying relationships between input variables and paddy yield in the dataset are predominantly linear. These findings highlight the importance of systematic model evaluation and demonstrate that simpler regression models can remain effective and interpretable for practical paddy yield prediction and agricultural decision support systems.
YOLO26n-Based Apple Leaf Disease Detection for Precision Agriculture Using Lightweight Deep Learning and Object Detection Darsiti Darsiti; Budiman; Dhika Wdiyanto; Tarsinah Sumarni
Bulletin of Intelligent Machines and Algorithms Vol. 1 No. 4 (2026): BIMA May 2026 Issue
Publisher : Maheswari Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.65780/bima.v1i4.22

Abstract

Early detection of apple leaf diseases is a critical factor in supporting agricultural productivity and minimizing losses caused by plant disease outbreaks. However, manual identification processes still have limitations in terms of accuracy, consistency, and time efficiency. This study aims to develop an apple leaf disease detection model based on object detection using YOLO26n to identify four main classes: Apple__BlackRot, Apple__CedarRust, Apple__Healthy, and Apple__Scab. The dataset was obtained from Kaggle in YOLO format, consisting of 2,754 training images and 687 validation images. The study employs a transfer learning approach with various data augmentation techniques, such as mosaic, mixup, copy-paste, rotation, translation, and HSV transformation, to enhance the model’s generalization ability. Evaluation was conducted using the Precision, Recall, mAP50, and mAP50-95 metrics. The results showed that the YOLO26n model achieved a Precision of 0.968, a Recall of 0.887, an mAP50 of 0.958, and an mAP50-95 of 0.880. The best performance was achieved on the Apple__BlackRot class with an mAP50-95 value of 0.987. The inference results also show that the model is capable of accurately localizing diseases through bounding boxes with a high level of confidence. These findings indicate that YOLO26n has great potential as an efficient and accurate lightweight model for the implementation of real-time precision agriculture-based plant disease detection systems.
An Analysis of the Impact of Zoom on Online Learning Using the Technology Acceptance Model Zatin Niqotaini; Budiman Budiman; Fahreja Ramadhan; Artika Arista; Esa Prakasa; Arafat Febriandirza; Nur Alamsyah; Rezza Novian Noor Rochmat; Henki Bayu Seta
Journal of Computing Innovations and Emerging Technologies Vol. 2 No. 1 (2026): Volume 2 No 1
Publisher : novamindpress

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.64472/jciet.v2i1.30

Abstract

This study aims to analyze the effect of using the ZOOM application at the University of Informatics and Business Indonesia (UNIBI) using the Technology Acceptance Model (TAM) approach, which is often used by some researchers to examine user acceptance of technology. This research is quantitative using descriptive method. The data analysis technique was carried out using SEM (Structural Equation Model) with AMOS (Analysis of Moment Structure) software. The population in this study were UNIBI students. Determination of the sample is carried out by proportional sampling, which is a proportional sampling method based on sub-populations. The results of this study prove that only 4 hypotheses are accepted from a total of 6 hypotheses proposed. The following is the percentage of the influence of each variable: a) Perceived Ease of Use (PEOU) is 28%, b) Perceived Usefulness (PU) is 74%, c) Attitude Toward Using (ATU) is 57%, d) Behavioral Intention to Use (BITU) is 65%, and e) Actual system usage (AU) is 75%. This proves that the use of the ZOOM application as an online learning medium cannot be fully explained by the Technology Acceptance Model.
Integration of base64 encoding, vigenère cipher, and LSB steganography for securing learning materials in learning management systems. Deden Pradeka; Deandy Zahran Vierdansyah; Devi Aprianti Rimadhani Agustini; Anugrah Adiwilaga; Syifa Aqila Humaira; Aidha Salsa Billa; Budiman
TEKNOSAINS : Jurnal Sains, Teknologi dan Informatika Vol 13 No 2 (2026): TEKNOSAINS: Jurnal Sains, Teknologi dan Informatika
Publisher : LPPMPK- Universitas Muhammadiyah Cileungsi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37373/tekno.v13i2.2101

Abstract

The increasing use of Learning Management Systems (LMS) in digital education raises significant concerns regarding the confidentiality and integrity of learning materials stored and transmitted within the platform. This study proposes a layered security mechanism that integrates cryptography and steganography to protect educational files in an LMS environment. The system implements Base64 encoding for data transformation, Vigenère cipher encryption for key-based protection, Fisher–Yates shuffle for ciphertext randomization, and Least Significant Bit (LSB) steganography to conceal encrypted data within digital images. The proposed method was implemented in a PHP-based LMS prototype and evaluated using learning files in .ppt and .pdf formats as test objects. Encrypted files were embedded into high-resolution cover images with a resolution of 6000 × 4000 pixels. Experimental results show that the system successfully performs the complete processes of encryption, embedding, extraction, and decryption with a 100% integrity success rate, indicating that the original files can be perfectly reconstructed after extraction. Image quality analysis shows high visual imperceptibility with PSNR values above 60 dB and very low MSE values. However, the embedding process introduces storage overhead, resulting in an average file size increase. These results demonstrate that the proposed layered approach effectively enhances data confidentiality and concealment in LMS environments, although optimization of storage efficiency remains necessary for large-scale implementationThe increasing use of Learning Management Systems (LMS) in digital education raises significant concerns regarding the confidentiality and integrity of learning materials stored and transmitted within the platform. This study proposes a layered security mechanism that integrates cryptography and steganography to protect educational files in an LMS environment. The system implements Base64 encoding for data transformation, Vigenère cipher encryption for key-based protection, Fisher–Yates shuffle for ciphertext randomization, and Least Significant Bit (LSB) steganography to conceal encrypted data within digital images. The proposed method was implemented in a PHP-based LMS prototype and evaluated using learning files in .ppt and .pdf formats as test objects. Encrypted files were embedded into high-resolution cover images with a resolution of 6000 × 4000 pixels. Experimental results show that the system successfully performs the complete processes of encryption, embedding, extraction, and decryption with a 100% integrity success rate, indicating that the original files can be perfectly reconstructed after extraction. Image quality analysis shows high visual imperceptibility with PSNR values above 60 dB and very low MSE values. However, the embedding process introduces storage overhead, resulting in an average file size increase. These results demonstrate that the proposed layered approach effectively enhances data confidentiality and concealment in LMS environments, although optimization of storage efficiency remains necessary for large-scale implementation
SHAP-ALE: A Novel Approach to Explainability in Mental Health Prediction using RFR Nur Alamsyah; Budiman Budiman; Wala Erpurini; Hani Fitria Rahmani
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3087

Abstract

The prediction of mental health disorders, such as depression, has become increasingly crucial as global mental health concerns continue to rise. In this study, the predictive task specifically focuses on estimating the prevalence of depressive disorders as the primary target variable, while other mental health conditions such as anxiety and schizophrenia are treated as explanatory features. While machine learning (ML) models, like Random Forest Regressor (RFR), offer high accuracy in predictions, their interpretability remains a challenge. This research introduces SHAP-ALE, an innovative hybrid explainability framework that integrates SHapley Additive exPlanations (SHAP) and Accumulated Local Effects (ALE) to address this gap. SHAP provides both global and local insights into feature contributions, while ALE visualizes feature-target relationships, mitigating bias caused by feature correlations. Using a dataset comprising various mental health disorders and demographic factors, the dataset used in this study was obtained from Kaggle and consists of 6,421 records covering multiple mental health disorder indicators and demographic attributes across different regions and years. RFR model demonstrated robust predictive performance with an R² score of 0.9984 and a Mean Squared Error (MSE) of 0.0016. SHAP analysis revealed that features such as schizophrenia and anxiety disorders significantly influenced predictions, while ALE identified nonlinear relationships between these features and depression prevalence. The combined insights from SHAP and ALE enhance the interpretability of the model, enabling better understanding of the complex factors underlying mental health disorders. This study highlights the contribution of SHAP-ALE as a hybrid explainability framework that integrates local and global interpretability, enabling a more comprehensive understanding of feature interactions and non-linear effects beyond the capabilities of individual methods.
A stacking ensemble model with SMOTE for improved imbalanced classification on credit data Nur Alamsyah; Budiman Budiman; Titan Parama Yoga; R. Yadi Rakhman Alamsyah
TELKOMNIKA (Telecommunication Computing Electronics and Control) Vol 22, No 3: June 2024
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/telkomnika.v22i3.25921

Abstract

This research is based on a significant problem in credit risk analysis in the banking sector caused by class imbalance. We face the problem of the model’s inability to accurately identify risks in the ‘‘Charged Off’’ class. As a solution, we propose a stacked ensemble approach that utilizes synthetic minority over-sampling technique (SMOTE) to balance the class distribution. Experiments were conducted by applying SMOTE to the training data before training the credit model using gradient boosting (XGBoost) and random forest (RF) algorithms in a single ensemble. The results show significant improvements in precision, recall, and F1-score after applying SMOTE on the unbalanced classes. The updated model achieved a striking accuracy rate of 0,97 on resampled training data. This re-search clearly identifies the problem of class imbalance as a major challenge in credit risk analysis. The application of SMOTE in a stacked ensemble was found to be effective in improving model performance, making a valuable contribution to the development of more reliable credit models for better risk management and revenue generation in financial institutions.
XGBoost optimization using hybrid Bayesian optimization and nested cross validation for calorie prediction Budiman Budiman; Nur Alamsyah; Titan Parama Yoga; R Yadi Rakhman Alamsyah; Elia Setiana
TELKOMNIKA (Telecommunication Computing Electronics and Control) Vol 23, No 3: June 2025
Publisher : Universitas Ahmad Dahlan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12928/telkomnika.v23i3.26554

Abstract

Accurately predicting calorie expenditure is crucial for wearable device applications, enabling personalized fitness and health recommendations. However, traditional models struggle with high data variability and nonlinear relationships in activity data, leading to suboptimal predictions. This study addresses these challenges by integrating extreme gradient boosting (XGBoost) with Bayesian optimization and nested cross validation to enhance predictive accuracy. Unlike previous approaches, our method systematically tunes hyperparameters using Bayesian optimization while employing nested cross validation to prevent overfitting, ensuring robust model evaluation. We utilize a dataset of daily activity records, including steps, distance, and active minutes, extracted from wearable devices. Our experimental findings indicate a substantial enhancement in prediction performance, achieving a mean squared error (MSE) of 4294.27, an Rsquared (R2) score of 0.9917, and a root mean squared error (RMSE) of 65.53. The proposed model outperforms baseline approaches such as random forest and support vector machines in terms of predictive accuracy. These findings underscore the advantage of our approach in predictive modeling. Beyond calorie estimation, the proposed methodology is adaptable to other domains requiring high-precision predictions, such as healthcare analytics and personalized recommendation systems.
ISOLATION FOREST PARAMETER TUNING FOR MOBILE APP ANOMALY DETECTION BASED ON PERMISSION REQUESTS Valencia Claudia Jennifer Kaunang; Nur Alamsyah; Reni Nursyanti; Budiman Budiman; Venia R Danestiara; Elia Setiana
Jurnal Pilar Nusa Mandiri Vol. 21 No. 2 (2025): Pilar Nusa Mandiri : Journal of Computing and Information System Publishing Pe
Publisher : LPPM Universitas Nusa Mandiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33480/pilar.v21i2.6647

Abstract

Ensuring mobile app security needs the capability to detect apps that request excessive or inappropriate permissions. This research proposes an anomaly detection approach using Isolation Forest, enhanced through hyperparameter tuning, to identify suspect apps based on permission request patterns. The dataset is processed into binary features, followed by exploratory data analysis (EDA) to examine the distribution and highlight sensitive permissions. The Isolation Forest model is then optimized by tuning parameters such as contamination level, number of estimators, and sample size. The fine-tuned model achieved a more accurate separation between normal and anomaly applications, detecting 10 anomalies out of 200 applications, with anomaly applications averaging 125.10 permits compared to 42.76 in normal applications. These anomalies often requested permissions related to network, storage, contacts and microphone, indicating potential privacy risks. The results show that parameter tuning improves the detection performance of Isolation Forest, providing a practical solution for mobile security monitoring. After tuning, the number of false positives decreased by 50%, and the model successfully reduced detected anomalies from 20 to 10, increasing the precision of anomaly detection from 70% to 90%. Future work could include improving feature selection and integration into real-time detection systems. 
A Progressive Training Framework for Robust YOLOv11-Based Vehicle Detection Across Domain Shifts in Real-World CCTV Environments Aditya Gunaldhi; Budiman; Chairul Habibi; Nur Alamsyah
Bulletin of Intelligent Machines and Algorithms Vol. 1 No. 5 (2026): BIMA July 2026 Issue
Publisher : Maheswari Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.65780/bima.v1i5.27

Abstract

The performance of deep learning-based vehicle detection models often deteriorates when applied to real-world CCTV environments due to domain shift caused by variations in lighting, occlusion, glare, and changes in camera viewpoint. This study aims to develop a YOLOv11-based progressive training framework to improve the model’s generalization ability under heterogeneous operational conditions. The proposed method consists of four stages: base training, fine-tuning, and a supervised progressive domain adaptation strategy implemented through sequential fine-tuning on labeled target-domain CCTV images before OpenVINO-based inference optimization. The model is trained using a source dataset and adapted to a target dataset representing real-world CCTV conditions. Evaluation was conducted using Precision, Recall, mAP@50, mAP@50–95, loss curve analysis, per-vehicle-class evaluation, visual testing on CCTV video, as well as latency and throughput measurements. The results show that base training achieved an mAP@50 of 0.919 and built a robust feature representation, while fine-tuning maintained performance stability with an mAP@50 of 0.915. Although domain adaptation reduced mAP@50 to 0.811 due to domain shift, the model demonstrated improved generalization capabilities and maintained more consistent detection under low-light conditions, glare, occlusion, and heavy traffic. OpenVINO INT8 optimization increased inference speed from 4.86 FPS to 9.08 FPS with minimal accuracy loss. These findings demonstrate that the progressive training framework effectively bridges differences in data distribution while producing a vehicle detection model that is more robust, efficient, and suitable for deployment in real-time CCTV-based traffic monitoring systems.
Co-Authors Acep Hendra Aditya Gunaldhi Aggi Panigoro Sarifiyono Ahmad Fauzi Ramadhan Aidha Salsa Billa Akbar, Imannudin Alamsyah, R Yadi Rakhman AlFauzi, Ihsan Alif Januantara Prima Almira Nurchawilah Amos Duan Nugroho Andi Sunyoto Angela Nisrina Zhafirama Anto Widianto Anugrah Adiwilaga Arafat Febriandirza Ari Rizki Fauzi Arief Setyanto Artika Arista Cahya Miftahul Falah Catherin Rumambo Mogot Pandin Chairul Habibi Chairul Habibi Chery Cardinawati Sitohang Danestiara, Venia R Dani Rizky Zaelani Darsiti Darsiti Deandy Zahran Vierdansyah Deden Pradeka Devi Aprianti Rimadhani Agustini Dhani Ariatmanto Dhika Wdiyanto Dirham Triyadi Dirham Triyadi Elia Setiana Encep Hamzah Esa Prakasa Fachrizal Ardiansyah Fahmi Abdullah Fahreja Ramadhan Fauzi Ramadhan, Ahmad Fikri Rizqillah Hasani Fitri Kinkin Gelar, Trisna Gunthur Bayu Wibisono Habibi, Chairul Hani Fitria Rahmani Hasan Nuraripin Henki Bayu Seta Hernawan, Kartika Nursyabanita Ihsan AlFauzi Ilham Ramadhan Ismi Nur Muhamad Jennifer Kaunang, Valencia Claudia Karlina, Nichi Hana Kaunang, Valencia Kaunang, Valencia Claudia Jennifer M. Rizki Ramadhan Muhammad Noerhadi Muhammad Rizki Ramadhan Nasution, Vani Maharani Niqotaini, Zatin Nur Alamsyah NUR ALAMSYAH Nur Alamsyah Nursyanti, Reni PARAMA YOGA, TITAN R Yadi Rakhman Alamsyah R. Yadi Rakhman A R. Yadi Rakhman A4 R. Yadi Rakhman Alamsyah R. Yadi Rakhman Alamsyah Raka Deny Abdi Putra Rakhman Alamsyah, Rd. Yadi Rd. Yadi Rakhman Alamsyah Rd. Zidni Rizan Al-Zhahir Yanuar Reni Nursyanti Reni Nursyanti Reni Nursyanti Reynaldy Gimnastiar Rezza Novian Noor Rochmat Rijwan Rijwan S.W. Manurip, Atanasius Angga Sardjono Setiana, Elia Silvana Anggraeni, Zulmeida Sophian Ramadhan Suci Fitriani Setiawan Syifa Aqila Humaira Tarsinah Sumarni Tiara Permata Hati Titan Parama Titan Parama Yoga Titan Parama Yoga Titan Parama Yoga Tutik Ultsa Rahmatika Valencia Claudia Jennifer Valencia Claudia Jennifer Kaunang Valencia Claudia Jennifer Kaunang Venia R Danestiara Venia Restreva Danestiara Wala Erpurini Wala Erpurini Wulandari Wulandari Yoga Rizki Rahmawan Yoga Rizki Rahmawan Zein Suna Arfigan Said