cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Ardiansyah
Contact Email
ardiansyah@fmipa.unila.ac.id
Phone
-
Journal Mail Official
jurnalpepadun@fmipa.unila.ac.id
Editorial Address
Gedung Ilmu Komputer Fakultas Matematika dan Ilmu Pengetahuan Alam - Universitas Lampung Jalan Soemantri Brojonegoro No.1 Bandarlampung
Location
Kota bandar lampung,
Lampung
INDONESIA
Jurnal Pepadun
Published by Universitas Lampung
ISSN : -     EISSN : 27743403     DOI : https://doi.org/10.23960/pepadun
Core Subject : Science,
Computer Science & IT
Pepadun Journal is a journal to publish research in the fields of computer science, information systems, and informatics to researchers, scientists, and professionals. For every edition published by the Pepadun Journal, we put our effort: Using standard procedures and times for submitted manuscripts, Provide a good editorial service for every submitted manuscript, Attract national and international writers to contribute to submitting quality manuscripts, Managing journals with good quality standards Pepadun is published three times a year by Computer Science Department, University of Lampung. Contributed papers must be original and offer a state-of-the-art contribution. Each manuscript will be peer-reviewed by reviewers in the relevant field ensuring the quality of the publication. Pepadun offers an open-access license (CC-BY), authors retain the copyright.
Arjuna Subject : Ilmu Komputer (semua kategori) - Aplikasi Ilmu Komputer
Articles 10 Documents
 1 
Search results for , issue "Vol. 7 No. 1 (2026): April" : 10 Documents clear
Application of MobileNetV1 and DenseNet-121 CNN Architectures for Eye-Based Gender Classification Nurhalifah, Sinta; Junaidi, Akmal; Kurniawan, Didik; Parabi, M. Iqbal
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i1.296

Abstract

Gender classification is an important field in biometric identification systems that plays a vital role in security, forensics, and human–computer interaction. Human eye images are a promising visual object for gender classification because they contain distinct anatomical features that differ between males and females. This study aims to implement and evaluate two Convolutional Neural Network (CNN) architectures, namely MobileNetV1 and DenseNet-121, for gender classification based on human eye images. The dataset used was obtained from the Kaggle platform, consisting of 11,525 eye images, with 6,323 male and 5,202 female samples. The research process involved several stages, including pre-processing, data splitting, augmentation using Affine transformations (rotation and translation), as well as model training and evaluation. The evaluation was conducted using accuracy, precision, recall, and F1-score metrics. The results showed that both architectures were capable of performing gender classification effectively, although differences in performance were observed. The best accuracy was achieved by MobileNetV1 with a rotation scenario of 92.49%, while DenseNet-121 obtained 86.84% with a combined rotation and translation scenario. This research is expected to contribute to the development of efficient and accurate eye image–based gender classification systems using deep learning approaches.
HYBE Corporation Stock Price Prediction Using CNN-LSTM with CRISP-DM Framework Hidayah, Nurul; Khaira, Ulfa; Bintana, Rizqa Raaiqa
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.297

Abstract

Digital transformation in financial analysis requires the application of computational models that can handle the complexity of the stock market efficiently and objectively. This study aims to predict the stock price of HYBE Corporation using the Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) model within the CRISP-DM (Cross Industry Standard Process for Data Mining) framework. The data used consists of daily stock prices of HYBE Corporation obtained from Yahoo Finance. The research process includes the main stages of CRISP-DM, namely business understanding, data understanding, data preparation, modeling, evaluation, and presentation of results. The CNN-LSTM model is designed to combine the ability of CNN to extract local patterns from time series with the advantage of LSTM in capturing long-term dependencies. To maximize the parameters used, this study also performed hyperparameter tuning using GridSearchCV on several key parameters. This process aimed to obtain the best combination of parameters capable of improving prediction accuracy and reducing the error value in the CNN-LSTM model. The evaluation results show that the CNN-LSTM model is capable of providing predictions with a very high level of accuracy. The Mean Squared Error (MSE) value is 0.00029, the Root Mean Squared Error (RMSE) is 0.01704, the Mean Absolute Error (MAE) is 0.01346, and the Mean Absolute Percentage Error (MAPE) is 2.15%. These low evaluation values demonstrate the model's effectiveness in handling stock market volatility while maintaining stability in predicting both short-term and long-term patterns.
Analysis of Product Purchase Patterns Using the Apriori Algorithm on FMCG Distributor Transaction Data in the Riau Region Muharmi, Yulya; Azwanti, Nurul; Amelia, Dhella
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.324

Abstract

This study investigates purchasing patterns of fast-moving consumer goods (FMCG) in Riau Province, Indonesia, using the Apriori algorithm within the Market Basket Analysis framework. Transaction data from a distributor comprising 4,422 transactions and 243 unique products across Pekanbaru, Kampar, and Rokan Hulu were analyzed to generate frequent itemsets and association rules, evaluated using support, confidence, and lift metrics. The application of a consistent minimum support and confidence threshold ensures statistically robust rule extraction across regions with different transaction scales.The results reveal strong intra-brand associations within the snack category, with several rules exhibiting lift values above ten, indicating systematic bundling behavior rather than random co-occurrence. These findings suggest that retailers tend to stock complementary product variants simultaneously, reflecting structured purchasing patterns at the outlet level. Regional comparison highlights differences in rule density across districts, shaped by transaction volume and the proportional effect of the support threshold, demonstrating how dataset scale influences association complexity. Overall, the study demonstrates that the Apriori algorithm effectively uncovers meaningful purchasing structures in distributor-level transaction data. The findings provide actionable insights for inventory management, regional distribution planning, and targeted promotions, while contributing to the literature by examining FMCG purchasing behavior in a multi-region distribution context using empirical distributor data.
Design and Implementation of an Arduino Uno Based Smart Automatic Water Dispenser Prototype Sakban, Muhammad; Hutabarat, Purnama Helena
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.332

Abstract

 This study aims to design and develop an Arduino Uno–based smart automatic dispenser as a hygienic solution for touchless liquid dispensing. Conventional dispensers require physical contact, increasing the risk of contamination; therefore, a contactless system is proposed to improve hygiene and user convenience. The system utilizes an Arduino Uno as the main controller, an HC-SR04 ultrasonic sensor for object detection, a relay module for switching control, and a DC motor to pump liquid. A 16×2 LCD is integrated to display real-time system status. Experimental testing was conducted using repeated trials to evaluate detection accuracy, response time, and error rate. The results show that the ultrasonic sensor achieves 100% accuracy at 5 cm and approximately 95% at 10 cm, with decreased performance at longer distances. The DC motor and relay system demonstrated a 90% success rate with an average response time of 1.5 seconds. Overall system testing achieved 95% accuracy (19 out of 20 successful trials). These results indicate that the system operates effectively and reliably under defined conditions and has potential for application in public facilities to support hygienic and touchless liquid dispensing.
Evaluating Kolmogorov-Arnold Networks for Multispectral Land Cover Classification Using Sentinel-2 Imagery in Jambi City Waladi, Akhiyar; Iftitah, Hasanatul
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.334

Abstract

Land cover maps obtained from satellite imagery are used in environmental management and spatial planning. Deep learning now outperforms traditional machine learning for this task, but Kolmogorov-Arnold Networks (KAN) have rarely been tested on multispectral remote sensing data. This paper evaluates two KAN strategies for classifying nine land cover types from Sentinel-2 imagery in Jambi, Indonesia. ResNet-KAN adds a KAN-based classifier head to a standard CNN backbone, while ConvKAN builds the entire network from KAN-based convolution layers. Both are compared against seven CNN, Transformer, and machine learning baselines using 23 spectral features with Google Dynamic World labels as reference, and ablation experiments test spectral feature composition, ImageNet transfer learning, and input patch size. Swin Transformer reaches the highest overall accuracy (88.34%), but ConvKAN better separates rare land cover classes like Grass and Shrub, achieving the best F1-Macro (0.5870) with only 2.91 million parameters, 89.4% fewer than Swin-T. Adding spectral indices raises ConvKAN F1-Macro by 13.8%, but lowers ViT accuracy by 3.19% OA because self-attention can already learn band-ratio operations from raw bands. KAN models also perform better when trained from scratch, because most Sentinel-2 channels fall outside the visible spectrum that ImageNet covers. Spatially, ConvKAN produces maps as clean as Swin Transformer despite being ten times smaller. KAN can therefore match larger models in accuracy and map quality for multispectral land cover classification.
Comparison of ARIMA and LSTM Models for Regional Export Values in Lampung Province Kurnia, Rian; Suciati, Indah; Nurmadani, Vina
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.342

Abstract

Accurate forecasting of regional export values is critical for effective macroeconomic planning. However, these indicators often exhibit complex volatility and structural shocks that challenge traditional frameworks. This study compares the forecasting performance of Autoregressive Integrated Moving Average (ARIMA) method against the machine learning architecture Long Short-Term Memory (LSTM), utilizing the monthly export values of Lampung Province. Data from January 2015 to December 2024 were partitioned into a training set (2015-2022) and testing set (2023-2024). For the linear approach, following Box-Cox transformation and first-order differencing, an ARIMA(1,1,0)(1,0,1)[12] model was fitted to the data based on Akaike Information Criterion (AIC) with comparison to other models of ARIMA. Simultaneously, an LSTM network was constructed using a 12-month lookback window and Min-Max scaling. The results indicate that the optimized ARIMA model achieved a lower Root Mean Squared Error (RMSE) of 94,030,344 compared to the LSTM network of 395,566,847 during the 24 months testing window. The ARIMA model effectively captured the underlying linear trends and stable annual seasonality without overfitting the training data. The study concludes that for moderately sized time series, ARIMA remains highly robust and can outperform complex machine learning architectures. Consequently, while neural networks offer advanced capabilities, classical frameworks should remain a primary tool for establishing baseline indicators in regional forecasting.  
Model Comparison and Feature Selection for Crop Recommendation Abdullah, Harnan Malik; Fahrurrozi, Imam
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i3.345

Abstract

Selecting crops appropriate to soil and environmental conditions is a crucial component of decision-making in smart agriculture. This study aims to evaluate the effect of feature selection on the predictive performance, stability, and computational efficiency of several machine learning models in crop recommendation tasks. The dataset employed in this study is a publicly available crop recommendation dataset, encompassing attributes such as nitrogen, phosphorus, potassium, temperature, humidity, pH, and rainfall. Six distinct models were evaluated: Logistic Regression, K-Nearest Neighbors, Support Vector Machine, Random Forest, XGBoost, and LightGBM, under two distinct conditions: utilizing all available features and employing a selection of features. The models' performance was assessed through various metrics, including accuracy, precision, recall, F1-score, the mean and standard deviation of cross-validation accuracy, as well as the inference time per sample.  Random Forest outperformed other models, achieving high accuracies (0.993–0.995) across both full and selected feature scenarios. The model input was simplified with minimal performance impact by the feature selection, which left the temperature and pH unselected. These results indicate that environmental factors, in addition to soil nutrients, substantially affect crop recommendations. Consequently, this research underscores that the evaluation of models for crop recommendation should prioritize not only accuracy but also stability, inference efficiency, and feature relevance to facilitate practical application within smart agricultural systems.
Implementation of the Haversine Algorithm in the Development of a WebGis Pambudi, Agung; Hanif, Ahmad Fikri; Rahmat, Syauqi
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i1.347

Abstract

Monitoring transmission tower disruptions require a system capable of presenting location information accurately and quickly to support the decision-making process. Limitations in the presentation of spatial data and the manual process of identifying disruption locations pose obstacles to improving the effectiveness of network infrastructure management. This study aims to develop a web-based transmission tower disruption monitoring system that integrates geospatial visualization and distance calculations to determine the nearest tower. The system is built using web technology with digital map support to display tower locations and disruption points interactively. Backend testing results using k6 show that all endpoints have a 100% success rate without failure, with response times ranging from 1.73 ms to 4430 ms, with the majority being below 1000 ms. All response values are still within the tolerance limit (<5000 ms), indicating stable system performance in handling various types of requests. In addition, the results of User Acceptance Testing (UAT) show that all system modules run according to user requirements.
A Scheduling Model for Balancing the Workload at Supermarket X Using Integer Programming Safika, Soni; Nurvazly, Dina Eka; Ansori, Muslim; Wamiliana, Wamiliana
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i1.348

Abstract

Employee scheduling is an important operational problem in the retail industry, as it affects both workforce efficiency and workload balance. This study aims to develop an employee scheduling model for Supermarket X using an integer programming approach based on simulated data. The model considers two work shifts, 42 employees, and an eight-day scheduling period, with decision variables representing work assignments and days off. Several operational constraints are included, such as minimum staffing requirements, one-shift-per-day restrictions, and limitations on the number of employees off on the same day. The model is solved using LINGO software. The results show that each employee is assigned seven working days and one rotating day off, while all constraints are satisfied. These results indicate that integer programming is an effective method for generating balanced and feasible employee schedules in retail operations.
Assessing Detection and Classification Performance for Vehicle License Plate Colors Using YOLOv5, YOLOv7, YOLOv8, and YOLOv9 Sholehurrohman, Ridho; Muhaqiqin, Muhaqiqin; Ilman, Igit Sabda; Habibi , Reza
Jurnal Pepadun Vol. 7 No. 1 (2026): April
Publisher : Department of Computer Science, Faculty of Mathematics and Natural Sciences, University of Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/pepadun.v7i1.349

Abstract

This study assesses the detection and classification performance of YOLOv5, YOLOv7, YOLOv8, and YOLOv9 for vehicle license plate colors in Indonesia, supporting the electronic ticketing system (e-tilang). The dataset consisted of 1,214 images from video footage captured in Bandar Lampung, comprising five color categories: black, white, yellow, red, and non-plate. The models were trained using transfer learning with COCO pre-trained weights, evaluated using precision, recall, F1-score, mAP50, and mAP50-95, and tested under real-world moderate and crowded traffic conditions. The results show that YOLOv9 consistently outperformed all other models, achieving the highest precision (97.20%), recall (96.50%), F1-score (96.85%), mAP50 (98.10%), and mAP50-95 (80.50%), with the fastest inference time of 6.8 ms per image (approximately 147 FPS). YOLOv8 ranked second, followed by YOLOv7 and YOLOv5. Across all models, the non-plate category remained the most challenging, while white and yellow plates were occasionally misclassified under low-light conditions. In conclusion, YOLOv9 is recommended for deployment in Indonesia's e-tilang system due to its best balance of accuracy and speed. Future work should expand the dataset to more diverse geographical locations, evaluate model performance under extreme weather conditions, and deploy the model on edge devices to validate real-world performance.

Page 1 of 1 | Total Record : 10

 1 

Filter by Year

2026 2026


Reset
Filter By Issues
All Issue Vol. 7 No. 1 (2026): April Vol. 4 No. 1 (2023): April Vol. 3 No. 3 (2022): December Vol. 3 No. 2 (2022): August Vol. 3 No. 1 (2022): April Vol. 2 No. 3 (2021): December Vol. 2 No. 2 (2021): August Vol. 2 No. 1 (2021): April Vol. 1 No. 1 (2020): December