ADA Journal of Information System Research

Pengembangan Sistem Informasi Mutasi Kendaraan Dinas (SIMUNDIS) Berbasis Web

Asti Sri Wahyuni — 2026-06-30

The official vehicle transfer process in Kecamatan Cibinong is still carried out manually using physical documents and personal communication. Submission data, transmission status, and Handover Report (BAST) documents are also not stored in a single system. This study aims to develop a web-based Official Vehicle Transfer Information System (SIMUNDIS) to manage the vehicle maintenance process from submission, inspection and verification, status monitoring, handover, BAST document management, to reporting. Data collection was carried out through observation, interviews, and literature studies. The system development adopted the Waterfall model, while the system was built using the CodeIgniter 4 framework and MySQL as a database. Testing was carried out using the Black Box Testing method with 8 scenarios covering login, vehicle data, transfer application, application confirmation, application details, and reports. The test results showed that 8 of the 8 scenarios were declared valid and ran according to the expected output, so that the test success rate reached 100%. The developed SIMUNDIS focuses on the official vehicle maintenance process between sub-districts and villages in Kecamatan Cibinong, including maintenance data, status applications, and BAST documents.

ChaosNet-Transformer: A Physics-Informed Deep Learning Framework for Predicting and Synchronizing High-Dimensional Chaotic Dynamical Systems

Wanayumini Wanayumini — 2026-06-30

Predicting and synchronizing high-dimensional chaotic dynamical systems remains a fundamental challenge in artificial intelligence and nonlinear science because chaotic trajectories exhibit strong sensitivity to initial conditions and rapid error divergence. This study proposes ChaosNet-Transformer (CNT), a physics-informed deep learning framework that combines Lyapunov-regularized temporal attention, Reservoir-Enhanced Embedding, and a Koopman Linearization Module to improve chaotic time-series prediction while preserving the underlying dynamical structure. CNT is evaluated on six benchmark systems, namely Lorenz-63, Rössler, Mackey-Glass, Chen-Lu, Qi Hyperchaos, and the Kuramoto-Sivashinsky system. The proposed model is compared with Echo State Network (ESN), Next-Generation Reservoir Computing (NG-RC), FEDformer, and Chaos-Informed LSTM (CI-LSTM) using Valid Prediction Time (VPT) measured in Lyapunov Time. Experimental results show that CNT achieves the highest VPT across all benchmark systems and consistently outperforms the strongest baseline methods. The synchronization experiment further demonstrates that CNT produces lower synchronization error than ESN and FEDformer under comparable coupling conditions. Ablation results confirm that Reservoir-Enhanced Embedding, Lyapunov regularization, and Koopman-based latent linearization each contribute to the model performance. These findings indicate that integrating Transformer-based sequence modeling with physics-informed constraints can improve the prediction and synchronization of chaotic dynamical systems.

Penerapan Random Forest untuk Klasifikasi Standar Gaji Industri Berdasarkan Level Pendidikan dan Pengalaman Kerja

Ramadhan Putra Prayoga — 2026-06-30

Determining salary compensation standards in industrial environments is often hindered by the high complexity of mapping multiple worker variables simultaneously, such as education level, work experience, age, and job title. Conventional determination processes that are not based on historical data correlations are prone to producing inconsistent and subjective decisions, triggering internal inequity. This study aims to design an objective, data-driven predictive model to automatically classify workers' salary standards into three target class categories (Low, Mid, High). The proposed computational solution is the application of Machine Learning methods using the Random Forest algorithm. The quantitative analysis phase was executed on 1,000 industrial worker profile records. To prevent data leakage phenomena that trigger pseudo-accuracy, the actual nominal salary attribute was strictly removed during the preprocessing phase. The system's performance evaluation was then validated using a 10-Fold Cross-Validation approach. The research results prove that the algorithm successfully extracted formal education and work experience variables as the most crucial primary splitting parameters. Based on parameter variation experiments, the model successfully reached its optimal learning convergence point at a configuration of 100 decision trees. At this point, the system recorded peak performance with an accuracy rate of 76.80%, a precision value of 0.767, and an F1-Score of 0.767. Modifying the computational structure by increasing it to 150 trees provided no further performance escalation. In conclusion, this analytical architecture is proven to be valid and reliable for implementation as a decision support instrument to establish an objective and transparent payroll system.