Claim Missing Document
Check
Articles

Found 10 Documents
Search

PELATIHAN DESIGN GRAFIS SEDERHANA DENGAN APLIKASI CANVA BAGI SISWA SMK NASIONAL DEPOK Kahfi Heryandi Suradiradja; Dani Ramdani; Slamet Raharjo
KOMMAS: Jurnal Pengabdian Kepada Masyarakat Vol 4, No 1 (2023): KOMMAS: JURNAL PENGABDIAN KEPADA MASYARAKAT
Publisher : KOMMAS: Jurnal Pengabdian Kepada Masyarakat

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

Abstract

Salah satu kewajiban yang dibebankan kepada dosen Universitas Pamulang adalah Pengabdian Kepada Masyarakat yang dilaksanakan setiap semester. Pada semester ganjil tahun ajaran 2022-2023, pengabdian dilaksanakan di SMK Nasional Depok dengan alasan pemilihan pemilihan tersebut karena salah satu SMK swasta yang memiliki fasilitas memadai dengan jumlah siswa yang lumayan banyak namun website yang dimilikinya masih perlu dikembangkan dalam hal inovasi konten yang ada pada websitenya, sehingga kemanfaatannya masih kurang maksimal sebagai media untuk menginformasikan profil, potensi, kegiatan, dan berbagai keunggulan yang dimiliki sekolah kepada masyarakat umum dengan desain grafis yang lebih menarik dan kreatif. Oleh karena itu tim dosen pengabdi mengusulkan suatu workshop atau pelatihan dengan tema Pemanfaatan Aplikasi Canva untuk Desain Grafis bagi Siswa SMK.  Pemilihan aplikasi tidak terlepas dari pertimbangan teknis siswa dan biaya sehingga dipilih aplikasi Canva berbasis web based baik digunakan dengan desktop komputer ataupun smartphone. Pelatihan ini diharapkan mampu menghasilkan siswa yang mampu membuat desain grafis atau konten grafis sederhana sehingga dapat mendukung kegiatan apapun yang positif yang dapat diberdayagunakan secara maksimal.
PENERAPAN NATURAL LANGUAGE PROCESSING ALGORITME SBERT PADA APLIKASI CHATBOT PENERIMAAN MAHASISWA BARU UNIVERSITAS PAMULANG Aji Sakti Ardiansyah; Kahfi Heryandi Suradiradja
Informatika: Jurnal Teknik Informatika dan Multimedia Vol. 6 No. 1 (2026): MEI : JURNAL INFORMATIKA DAN MULTIMEDIA
Publisher : LPPM Politeknik Pratama Kendal

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51903/informatika.v6i1.1645

Abstract

The New Student Admission (PMB) process at Pamulang University currently relies on one-way information channels such as Instagram and brochures only at designated posts. This condition limits access to information for prospective students, especially when they require quick and detailed answers regarding registration requirements, schedules, and tuition fees. Furthermore, the large number of applicants in each admission period makes manual inquiry services inefficient and prone to delays. Therefore, a technology-based solution must provide interactive, fast, and accurate information services. This study aims to develop a Chatbot based on Natural Language Processing (NLP) using the Sentence-BERT (SBERT) algorithm to support the PMB process at the Informatics Engineering Program at Pamulang University. The Chatbot is designed to provide accurate and relevant automated responses to prospective students' inquiries regarding PMB. The SBERT algorithm enhances the Chatbot’s ability to understand sentence meanings and deliver contextual responses. The software development adopts the Waterfall model, encompassing requirements analysis, system design, implementation, and testing phases. The expected outcome is a Chatbot system with- high response accuracy that improves efficiency and accessibility of PMB information services. This research is expected to contribute to the implementation of NLP technology in academic environments.
Analisis Prediksi Permintaan Produk FMCG menggunakan Model LSTM dan DES Untuk Optimalisasi Operasional Gudang Distribusi Tanti Cahya Herdiyani; Taswanda Taryo; Kahfi Heryandi Suradiradja; Zafira Salsabilah
BIOS : Jurnal Teknologi Informasi dan Rekayasa Komputer Vol 7 No 2 (2026): September
Publisher : Puslitbang Sinergis Asa Professional

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37148/bios.v7i2.213

Abstract

The Fast-Moving Consumer Goods (FMCG) sector is characterized by rapid product turnover and short shelf lives, requiring effective inventory management. Companies such as PT Macrosentra Niagaboga face challenges in maintaining stock availability. Although the inventory system has been integrated, stock verification is still conducted manually, requiring adaptive management to minimize the risk of overstock and stockout. Therefore, this study aims to develop a demand forecasting model using the Long Short-Term Memory (LSTM) algorithm integrated with Discrete-Event Simulation (DES) to optimize inventory management. The study utilized historical shipment data from May 2025 to January 2026. Preprocessing included data cleansing, date validation, and aggregation into daily and weekly data. Forecasting results were subsequently used as inputs for the DES simulations to determine optimal inventory policies. The results demonstrated that weekly LSTM aggregation was more accurate and stable than daily aggregation, as evidenced by a reduction in WAPE from 49.9% to 16.94% for high-demand products. Furthermore, integration with DES reduced stock levels by more than 30% using a safety stock ratio of 0.7 without compromising service levels. Finally, the proposed model was implemented in a web-based dashboard serving as a decision support system for monitoring forecasting and inventory policy recommendations.
Komparasi Decision-Tree dan Naïve Bayes untuk Deteksi Hama dan Penyakit Tanaman Anggur Akhmad Murtado; Kahfi Heryandi Suradiradja
RIGGS: Journal of Artificial Intelligence and Digital Business Vol. 5 No. 2 (2026): Mei-Juli
Publisher : Prodi Bisnis Digital Universitas Pahlawan Tuanku Tambusai

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31004/riggs.v5i2.12347

Abstract

Kebun Ruang Edukasi Literasi (REL) Rawa Buntu menghadapi kendala dalam mengidentifikasi hama dan penyakit tanaman anggur karena proses deteksi yang masih dilakukan secara manual berdasarkan pengalaman subjektif. Hal ini memicu risiko ketidakakuratan diagnosis yang berdampak pada kegagalan panen. Penelitian ini bertujuan untuk membuat aplikasi deteksi hama dan penyakit tanaman anggur dengan model terbaik dari hasil komparasi antara algoritme Decision Tree dan algoritme Naïve Bayes. Tahapan pada metodelogi penelitian ini meliputi pengumpulan data, pra-proses, pemodelan, evaluasi dan penerapan model. Dataset dari pengumpulan data diambil dari data histori pencatatan terhadap tanaman anggur di Kebun Ruang Edukasi Literasi Rawabuntu. Pemodelan dilakukan menggunakan tool RapidMiner dan evaluasi pengujian model menggunakan confusion matrix. Hasilnya menunjukkan bahwa algoritme Decision Tree menghasilkan performa terbaik dengan tingkat akurasi mencapai 94,12%. Tahap implementasi model dilakukan dengan menerapkan hasil aturan keputusan (rule IF-THEN) yang diekstraksi dari model algoritme Decision Tree pada sebuah aplikasi berbasis web untuk mendeteksi hama dan penyakit pada tanaman anggur di Kebun Ruang Edukasi Literasi Rawabuntu sehingga dapat membantu pengelola kebun dalam mendiagnosis gangguan tanaman anggur secara cepat, tepat, dan akurat. Saran untuk penelitian selanjutnya adalah jumlah dataset yang digunakan lebih banyak dan bervariasi lagi untuk kondisi gejala penyakit anggur sebagai input dari pemodelan.
Implementasi Random Forest Regression untuk Prediksi Kebutuhan Bahan Baku Minyak Senggugu Muhamad Nabukatneja Dwi Kusnandar; Kahfi Heryandi Suradiradja
Jurnal Komputer Teknologi Informasi Sistem Komputer (JUKTISI) Vol. 5 No. 2 (2026): September 2026
Publisher : LKP KARYA PRIMA KURSUS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62712/juktisi.v5i2.1572

Abstract

CV Tujuh Putik merupakan perusahaan yang memproduksi minyak senggugu dan menghadapi kesulitan dalam merencanakan kebutuhan bahan baku secara akurat karena proses perencanaan masih dilakukan secara manual berdasarkan perkiraan. Kondisi tersebut berpotensi menimbulkan kelebihan maupun kekurangan stok yang dapat mengganggu kelancaran produksi. Penelitian ini bertujuan untuk membangun model prediksi kebutuhan bahan baku minyak senggugu menggunakan tiga algoritme machine learning, yaitu Linear Regression, Support Vector Regression (SVR), dan Random Forest Regression, serta mengimplementasikan Random Forest Regression ke dalam prototipe aplikasi berbasis web. Metodologi penelitian menggunakan tahapan CRISP-DM yang meliputi pemahaman bisnis, pemahaman data, praproses data, pemodelan, evaluasi, dan deployment. Dataset yang digunakan berasal dari data historis operasional CV Tujuh Putik dengan variabel input berupa event e-commerce, kegiatan praktek, produk laku, bulan, dan hari, sedangkan variabel target berupa jumlah bahan baku terpakai. Dataset dibagi menjadi 80% data latih dan 20% data uji, kemudian dievaluasi menggunakan Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), dan R² Score. Hasil pengujian menunjukkan bahwa Support Vector Regression memperoleh nilai MAE dan RMSE terendah, masing-masing sebesar 0,0845 dan 0,1485, sedangkan Linear Regression memperoleh R² Score tertinggi sebesar 0,9958. Random Forest Regression menghasilkan MAE sebesar 0,1345, RMSE sebesar 0,3896, dan R² Score sebesar 0,9907. Berdasarkan fokus penelitian, Random Forest Regression kemudian diimplementasikan sebagai model utama pada prototipe aplikasi berbasis Flask. Aplikasi mampu menghasilkan estimasi kebutuhan bahan baku berdasarkan masukan pengguna. Pengujian sistem dilakukan menggunakan black box dan white box testing untuk memeriksa fungsi serta alur logika aplikasi.
Digital Forensic Analysis of Signature Images Using Error Level Analysis, Image Hashing, and Support Vector Machine Within the DFRWS Framework Amelia Yahya; Taswanda Taryo; Kahfi Heryandi Suradiradja
International Journal Software Engineering and Computer Science (IJSECS) Vol. 6 No. 2 (2026): AUGUST 2026
Publisher : Lembaga Komunitas Informasi Teknologi Aceh (KITA), Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35870/ijsecs.v6i2.7224

Abstract

The increasing use of digital documents in administrative and legal activities has expanded the use of image-based signatures for authentication and verification. However, signature images are vulnerable to manipulation using image-editing software, potentially resulting in document forgery and disputes over authenticity. This study examined the use of Error Level Analysis (ELA), perceptual hashing (pHash), and the Gray Level Co-occurrence Matrix (GLCM) to detect manipulation in signature images. It also evaluated the performance of a Support Vector Machine (SVM) in classifying genuine and forged signatures within the Digital Forensic Research Workshop (DFRWS) framework. The dataset comprised 720 signature images obtained from the Starter Handwritten Signatures Dataset. The research process involved image preprocessing, feature extraction, model training, and performance evaluation using a confusion matrix, accuracy, precision, recall, and F1-score. The model achieved an accuracy of 80.56% on previously unseen test data. The developed system also produced visual analysis outputs and generated digital investigation reports based on the DFRWS framework. These results indicate that the combination of ELA, pHash, GLCM, and SVM can support a structured digital forensic process for distinguishing between genuine and forged signature images.
Pengembangan Model Super Resolution Lightweight Berbasis Hybrid untuk Lingkungan Komputasi Terbatas Ade Soekarno Putra Santoso; Kahfi Heryandi Suradiradja; Choirul Basir
Journal of Creative Student Research Vol. 4 No. 4 (2026): Agustus: Journal of Creative Student Research
Publisher : Lembaga Pengembangan Kinerja Dosen

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55606/jcsr-politama.v4i4.6537

Abstract

Computational resource limitations, particularly the use of CPU-based systems without GPU acceleration, require super-resolution models that can achieve a balance between reconstruction quality and computational efficiency. This study proposes a lightweight hybrid super-resolution model designed for resource-constrained computing environments. The proposed architecture integrates lightweight feature extraction, residual learning, and an efficient image reconstruction module to enhance image resolution while maintaining low computational complexity. Model performance is evaluated using Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), computation time, CPU utilization, and memory consumption under different scaling factors. Experimental results show that the proposed model achieves competitive reconstruction performance while maintaining computational efficiency compared with several lightweight benchmark models. These findings indicate that the proposed hybrid architecture is suitable for deployment in resource-limited computing environments. Future work may focus on evaluating and optimizing the model on edge computing platforms, including smartphones and embedded systems, to validate its performance under real-world operating conditions.
Comparison of ConvLSTM, ConvLSTM with Temporal Embedding, and 3D CNN for Monthly Rainfall Prediction in Bekasi and Surroundings Areas Niken Wahyuni; Kahfi Heryandi Suradiradja; Ahmad Musyafa
SISTEMASI Vol 15, No 10 (2026): Sistemasi: Jurnal Sistem Informasi
Publisher : Universitas Islam Indragiri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32520/stmsi.v15i10.6968

Abstract

Monthly rainfall in tropical regions exhibits complex spatiotemporal variability, requiring models to simultaneously capture spatial heterogeneity and temporal patterns. This study compares ConvLSTM, ConvLSTM with Temporal Embedding (ConvLSTM-TE), and 3D CNN for monthly rainfall prediction using CHIRPS v2.0 data (1990–2025) over a 20 × 20 grid domain (5.75°–6.75° S; 106.5°–107.5° E) covering Bekasi and its surrounding areas. Twelve monthly rainfall maps were used to predict the subsequent month's rainfall map, with consistent temporal splits, forecast horizons, and evaluation metrics across all models. Hyperparameter tuning covered the number of filters, learning rate, batch size, scaling methods, and recurrent dropout, the latter being tested only for recurrent models. ConvLSTM-TE incorporates a 16-dimensional calendar representation derived from month and season embeddings, along with sine–cosine features. Climatology and persistence were used as baseline benchmarks. Across the 400-grid domain, the best ConvLSTM-TE configuration, selected based on test-set RMSE, achieved an RMSE of 82.368 mm/month, an MAE of 59.024 mm/month, a correlation coefficient (r) of 0.868, an R² of 75.25%, and an RSR of 0.497. It was followed by ConvLSTM (RMSE: 86.048 mm/month) and 3D CNN (RMSE: 96.347 mm/month). All three models outperformed persistence (RMSE: 125.639 mm/month), but only the two ConvLSTM-based models outperformed climatology (RMSE: 88.336 mm/month), achieving RMSE skill scores of 6.8% for ConvLSTM-TE and 2.6% for ConvLSTM. ConvLSTM-TE also achieved the lowest median RMSE across 92 training runs and retained the lowest RMSE under validation-loss-based model selection, although its performance advantage over ConvLSTM was marginal. The models preserved spatial gradients and seasonal phases but tended to smooth extreme rainfall events. An independent evaluation of ConvLSTM-TE against CHIRPS data for January–June 2026 yielded a mean absolute error of 57.15 mm/month, comparable to its test-set performance. The selected model was implemented in a Streamlit prototype for visualization.
Implementation of the Random Forest Regression Algorithm for Estimating the Number of Personnel at the Tangerang City Fire Department Call Center Rahel Azzahra; Kahfi Heryandi Suradiradja
Jurnal Media Computer Science Vol 5 No 4 (2026): Oktober
Publisher : LPPJPHKI Universitas Dehasen Bengkulu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37676/jmcs.v5i4.12568

Abstract

This study aims to apply the Random Forest Regression algorithm to estimate the number of personnel required for the Tangerang City Fire Department Call Center, based on historical fire incident data from 2023 to 2025. The research addresses the need for rapid, accurate, and data-driven personnel estimation to support decision-making by personnel. The study employs the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology, comprising Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Understanding. The variables utilized include year, time of incident, district, object type, fire severity level, weather, fleet size, and response time, with the number of personnel serving as the target variable. The study compares Random Forest Regression with Linear Regression using MAE, MSE, RMSE, and R² metrics. Test results indicate that Random Forest Regression achieved an MAE of 1.976, MSE of 6.085, RMSE of 2.467, and R² of 0.953, demonstrating superior performance compared to Linear Regression. The model was subsequently implemented in a Flask-based web application to serve as an operational tool for relevant personnel
Application of a 1D-CNN Model for Flood Forecasting Using Weather Parameters and Flood History in the Tangerang Region Nely Ramah Kurniawati; Kahfi Heryandi Suradiradja; Choirul Basir
Bulletin of Network Engineer and Informatics Vol. 4 No. 2 (2026): BUFNETS (Bulletin of Network Engineer and Informatics) October 2026
Publisher : PT. GWEX NET PUBLISHER

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59688/cvxgp523

Abstract

Flooding frequently affects the Tangerang region and causes substantial social and economic losses. Conventional forecasting methods may not adequately capture the nonlinear relationships between dynamic urban weather conditions and flood events. This study applies a one-dimensional convolutional neural network (1D-CNN) to classify flood potential using multivariate weather parameters and historical flood-event records. Temperature, relative humidity, and rainfall observations from three Automatic Weather Stations were combined with flood-event data from 2022 to 2024. Preprocessing included quality-control checks, outlier treatment, Multiple Imputation by Chained Equations (MICE), moving-average and lag features, sliding-window sequence construction, Min-Max normalization, and an 80:20 training-testing split. The model used 40 convolutional filters, max pooling, a 40-neuron hidden layer, and a three-class Softmax output representing low, medium, and high flood potential. The 1D-CNN achieved 83.59% accuracy, 83.52% precision, 83.59% recall, and an F1-score of 83.52%. The confusion matrix showed strong identification of low- and high-risk cases, although the medium class and some high-risk events remained difficult to distinguish. These results indicate that integrating weather observations with historical disaster records can support the development of a data-driven flood early warning framework for the Tangerang region.