Claim Missing Document
Check
Articles

IMPLEMENTASI METODE YOU ONLY LOOK ONCE (YOLOv5) DALAM DETEKSI PELANGGARAN HELM Meidyan, Martinus Ade; Yustanti, Wiyli
Journal of Emerging Information Systems and Business Intelligence (JEISBI) Vol. 5 No. 3 (2024)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jeisbi.v5i3.60517

Abstract

Pelanggaran pada lalu lintas yang di sebabkan oleh pengendara roda dua yang sering di tindak pada saat melakukan operasi patuh pada tahun 2023 mencatat, terdapat tiga pelanggaran terbanyak yang dilakukan oleh kendaraan roda dua. Paling banyak adalah pelanggaran tidak menggunakan helm yaitu sebanyak 8.916 pelanggaran (Made et al., 2020). Tujuan penelitian ini adalah menghasilkan sistem deteksi kendaraan berdasarkan kelasnya melalui analisis video berbasis algoritma YOLOv5. Metode yang disajikan dalam penelitian ini berfokus pada optimasi dan implementasi algoritma YOLOv5 untuk mendeteksi objek berupa helm pada pengendara roda dua pada saat berkendara, menggunakan dataset berisi 2000 gambar, dengan 1200 gambar untuk pelatihan dan 800 gambar untuk pengujian. Pelatihan dilakukan hingga mencapai langkah 200 epoch dengan batch 48 dengan ukuran gambar 448. Hasil penelitian dan uji coba berdasarkan eksperimen yang penulis lakukan, penulis berhasil mencapai nilai F1 Score sebesar 0.87 dan nilai mAP 0.90 menggunakan algoritma YOLOv5 dengan arsitektur YOLOv5m. Adanya beberapa faktor yang memengaruhi hasil deteksi adalah latar belakang objek pada gambar, posisi objek, terdapat objek penghalang pada sudut tertentu, serta tinggi/jarak objek.
Cat Skin Disease Detection System Using You Only Look Once (YOLO) v8 Algorithm: Sistem Deteksi Penyakit Kulit Kucing Menggunakan Algoritma You Only Look Once (YOLO) v8 meilita, Bunga; Yustanti, Wiyli
Journal of Emerging Information Systems and Business Intelligence (JEISBI) Vol. 5 No. 2 (2024)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jeisbi.v5i2.60656

Abstract

Kucing adalah hewan peliharaan yang popular di Indonesia, dengan jumlah populasi mencapai 4,80 juta ekor pada tahun 2022. Meskipun menggemaskan dan menyenangkan, kucing rentan terkena penyakit, terutama penyakit kulit speerti jamur. Pemilik hewan masih banyak yang kurang memahami gejala penyakit kulit kucing, sehingga penanganan penyakit tidak tepat yang bisa memperparah kondisi kucing. Solusi untuk mengatasi permasalah tersebut dengan mengimplementasikan algoritma You Only Look Once (YOLO) v8 yang dapat dijalankan secara realtime untuk mendeteksi penyakit kulit kucing jamu, scabies, lain dan sehat melalui aplikasi android. Berdasarkan hasil uji didapatkan Map score sebesar 0.788, precission sebesar 0.727, recall sebesar 0.769, dan F1-Score sebesar 0.75. Hasil pengujian white box berhasil berjalan pada semua test case yang ada. Hasil blackbox testing yaitu aplikasi bisa berjalan sesuai yang diharapkan, selain itu hasil uji pada fitur camera detector dengan pengujian ditiga jarak yang berbeda didapatkan jarak yang paling optimal untuk melakukan pendeteksian penyakit kulit kucing secara real time yaitu 20 cm dengan akurasi pengujian sebesar 0.92. Hasil uji pada fitur import gambar menghasilkan keakuratan sebesar 0.92.
Peramalan Jumlah Incident Information Technology PT XYZ Menggunakan Artificial Neural Network (ANN): Peramalan Jumlah Incident Information Technology PT XYZ Menggunakan Artificial Neural Network (ANN) Amalia, Dini; Yustanti, Wiyli
Journal of Emerging Information Systems and Business Intelligence (JEISBI) Vol. 5 No. 3 (2024)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jeisbi.v5i3.61037

Abstract

Forecasting is a technique for predicting events that will occur in the future using historical data as a comparison In this research, researchers try to determine the performance of the ANN method for forecasting the number of incidents at PT XYZ and build an application for forecasting the number of incidents at PT XYZ. This research uses the Mean Absolute Percentage Error (MAPE) evaluation metric as the evaluation metric that will be interpreted. The smaller the MAPE value, the better the model architecture. The best model is the model that produces the smallest MAPE value and does not experience underfitting or overfitting conditions. Based on the research results, it was found that all the best models from each model architecture produced a MAPE value of less than 10 and did not experience underfitting or overfitting conditions. Therefore, it can be interpreted that all the models produced are very accurate to be used as incident forecasting models at PT XYZ for the next 4 weeks. The website-based incident forecasting application created to predict the number of incidents for the next 4 weeks using the best model that has been previously saved also produces a MAPE value of less than 10 and does not experience underfitting and overfitting conditions.
Implementation of the Support Vector Machine (SVM) Algorithm in Predicting Transaction Cancellations at Shopee E-commerce: Implementasi Algoritma Support Vector Machine (SVM) Dalam Memprediksi Pembatalan Transaksi Pada E-commerce Shopee Maulidia, Ridhotul; Yustanti, Wiyli
Journal of Emerging Information Systems and Business Intelligence (JEISBI) Vol. 6 No. 1 (2025)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jeisbi.v6i1.64414

Abstract

In the digital era, shopping through e-commerce such as Shopee has become increasingly popular. However, transaction cancellation is still an obstacle that causes financial losses for sellers. This research aims to predict transaction cancellation on the Shopee platform using the Support Vector Machine (SVM) algorithm, which is expected to help sellers reduce the risk of loss. The data used comes from the transaction history of Shopee store nafystore.id and is processed using the CRISP-DM method, including business understanding, data preparation, modeling, and deployment. The data preparation process includes cleaning, encoding, normalization, and dimension reduction using Principal Component Analysis (PCA), as well as handling data imbalance with SMOTE. Model testing was conducted using K-Fold Cross-Validation at 3, 5, and 10 folds with different SVM kernels, where the linear kernel showed the best performance with 95.57% accuracy, 95.96% precision, 95.57% recall, and 95.58% F1-Score. The implementation of a web-based system is done using Streamlit to make it easier to use for sellers. The results of this research provide benefits for sellers in identifying cancellation factors, such as Total Payment and Estimated Shipping Fee Deductions. This research not only enriches the application of SVM algorithm in e-commerce analysis, but also provides a reference for other e-commerce platforms to improve transaction efficiency and customer satisfaction.
Hybrid Clustering and Classification of At-Risk Customer Segments in Network Marketing Hartanto, Unung Istopo; Buditjahjanto, I Gusti Putu Asto; Yustanti, Wiyli
JIEET (Journal of Information Engineering and Educational Technology) Vol. 9 No. 1 (2025)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jieet.v9n1.p42-50

Abstract

Customer segmentation is a fundamental strategy for sustaining retention in network marketing businesses, where repeated transactions and multilayered relationships significantly impact long-term customer value. This study proposes a hybrid machine learning framework to classify at-risk customer segments—comprising regular customers, seasonal buyers, and churn-risk profiles—by integrating unsupervised clustering and supervised classification methods. A total of 36 engineered behavioral features were derived from longitudinal transaction data to capture spending behavior, recency, variability, and growth dynamics. Clustering algorithms including K-Means, Agglomerative Hierarchical Clustering, and Gaussian Mixture Models were applied and evaluated using standard clustering validity indices: Silhouette Score, Davies–Bouldin Index, and Calinski–Harabasz Index. K-Means with six clusters produced the most interpretable and balanced segmentation outcome. Cluster relabeling was conducted to align with business-relevant categories, followed by supervised validation using classifiers such as Decision Tree, Gradient Boosting, K-Nearest Neighbors (KNN), Random Forest and Support Vector Machine (SVM). Among these, SVM yielded the highest predictive accuracy (92.53%) and F1-Score (92.52). The results demonstrate the effectiveness of the proposed hybrid approach in enhancing segmentation precision and facilitating early detection of potential churn in a dynamic marketing environment.
Automated Chest X-Ray Captioning Using Pretrained Vision Transformer with LSTM and Multi-Head Attention Aulia Akbar, Rafy; Putra, Ricky Eka; Yustanti, Wiyli
JIEET (Journal of Information Engineering and Educational Technology) Vol. 9 No. 1 (2025)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jieet.v9n1.p1-10

Abstract

Radiology report generation is a complex and error-prone task, especially for radiologists with limited experience. To overcome this, this study aims to develop an automated system for generating text-based radiology reports using chest X-ray images. The proposed approach combines computer vision and natural language processing through an encoder-decoder architecture. As an encoder, a Vision Transformer (ViT) model trained on the CheXpert dataset is used to extract visual features from X-ray images after Gamma Correction is performed to improve image quality. In the decoder section, word embeddings from the report text are processed using Long Short-Term Memory (LSTM) to capture word order relationships, and enriched with Multi-Head Attention (MHA) to pay attention to important parts of the text. Visual and text features are then combined and passed to a dense layer to generate text-based radiology reports. The evaluation results show that the proposed model achieves a ROUGE-L score of 0.385, outperforming previous models. The BLEU-1 score also shows competitive results with a value of 0.427. This study shows that the use of pre-trained ViT, combined with LSTM-MHA on the decoder, provides excellent performance in capturing visual and semantic context of text, as well as improving accuracy and efficiency in radiology report automation.
I-Regs (Internet-Regression Analysis) as a Statistical Innovation in Nonparametric Regression Modeling Dani, Andrea; Budiantara, I Nyoman; Nuraini, Ulfa Siti; Yustanti, Wiyli; Sifriyani; Putra, Fachrian Bimantoro
Journal of Education Technology and Information System Vol. 1 No. 02 (2025): Journal of Education Technology and Information System (JETIS)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jetis.v1i02.35288

Abstract

This research develops an information system based on the R-Shiny Dashboard, allowing users to perform nonparametric regression modeling. Internet-Regression Analysis (I-Regs) is the name of a dashboard that has been successfully developed. I-Regs provides a complete model library in regression analysis modeling, including parametric, nonparametric, and semiparametric regression. It is hoped that I-Regs can become a valuable tool for researchers, practitioners, and students in modeling regression analysis and solving various data analysis problems.
Customer Profiling and Purchase Patterns Using K-Means and Apriori Algorithms Muhammad Risalah Naufal; Yustanti, Wiyli
Journal of Education Technology and Information System Vol. 3 No. 01 (2027): Journal of Education Technology and Information System (JETIS)
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This study aims to analyze customer segmentation and purchasing patterns at PT. Benteng Api Technic (BAT) using the K-Means and Apriori algorithms. Customer segmentation is based on the RFM (Recency, Frequency, Monetary) approach, which reflects customer purchasing behavior. The K-Means algorithm is applied to group customers into clusters with similar characteristics, while the Apriori algorithm is used to identify frequent product purchasing patterns within each cluster. The dataset used consists of sales transaction data from June 1, 2023, to June 30, 2024. The results show clear customer segmentation based on purchasing characteristics, and several associations between products frequently purchased together by customers in specific clusters were found. These findings are expected to help the company develop more targeted marketing strategies and improve inventory management efficiency.
A COMPARATIVE STUDY OF SUPERVISED FEATURE SELECTION METHODS FOR PREDICTING UANG KULIAH TUNGGAL (UKT) GROUPS Putri, Windy Chikita Cornia; Yustanti, Wiyli; Yohannes, Ervin
J-Icon : Jurnal Komputer dan Informatika Vol 13 No 2 (2025): October 2025
Publisher : Universitas Nusa Cendana

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35508/jicon.v13i2.23893

Abstract

The manual classification of Uang Kuliah Tunggal (UKT) groups at Indonesian public universities is laborious, subjective, and error-prone, especially given the explosion of socio-economic data captured via online admission portals. In this study, we evaluate five feature selection techniques Chi-Square filter, Random Forest importance, Recursive Feature Elimination, LASSO embedded selection, and Exploratory Factor Analysis on a dataset of 9,369 applicants described by 53 socio-economic variables. Six classifiers (Decision Tree, Random Forest, SVM-RBF, K-Nearest Neighbor, and Naïve Bayes) were tuned via stratified 5-fold cross-validation within an 80:20 train-test split. Performance was measured by accuracy, macro-F1, and training time, and differences in weighted-average accuracy across feature-selection scenarios were assessed using the Friedman test (χ² = 15.06, p = 0.010). Results show that reducing to 13 features via LASSO (weighted-average accuracy 0.730) or Chi-Square (0.678) significantly outperforms both the full feature baseline (0.624) and the EFA baseline (0.303), while cutting computational costs by over 40%. We conclude that supervised feature selection particularly LASSO and Chi-Square enables simpler, faster, and more transparent UKT prediction without sacrificing accuracy. The novelty of this study lies in comparing five feature-selection methods within a standardized preprocessing pipeline on real UKT data from UNESA, resulting in a 13-feature subset aligned with the current UKT policy. This finding is ready to be integrated into an automated UKT verification system to enhance decision accuracy and efficiency.
Comparison You Only Look Once (Yolo) Algorithm On Physical Violence Video Detection Rahayu, Aulia Anisa Puji; Yustanti, Wiyli
Journal of Emerging Information Systems and Business Intelligence (JEISBI) Vol. 6 No. 3 (2025): Vol. 06 Issue 03
Publisher : Universitas Negeri Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26740/jeisbi.v6i3.71185

Abstract

Physical violence is one of the crimes that often occurs in various environments and can have a serious impact on victims, both physically and mentally. One of the obstacles in handling it is the delay in detecting acts of violence. The solution to this problem is to implement the best algorithm between You Only Look Once (YOLO) version 8 and version 9 to detect physical violence through video automatically and quickly. The dataset used consists of two classes, namely violence and non-violence, which have gone through the process of extraction, data cleaning, and labeling using Roboflow. The model was trained using Google Collaboratory, and the training results were evaluated using mAP, precision, recall, and F1-score metrics. Based on the test results, YOLOv9 obtained the best performance with a precision of 0.8096, recall of 0.8665, F1-score of 0.8363, and mAP of 0.8117. The detection system is then implemented into a web-based application using the Flask framework, which allows users to Upload videos and detect acts of violence automatically. The test results show that the application runs according to its function and is able to detect physical violence well. This research is expected to be a supporting solution in video-based security surveillance systems.