Jurnal Sekolah Tinggi Teknik Surabaya
Not a member yet
194 research outputs found
Sort by
Comparative Analysis of Large Red Chili Price Forecasting Models in Malang Regency Using Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA)
Full text linkLarge red chili is a strategic food commodity with high demand, yet its price often fluctuates due to factors such as weather, harvest seasons, and market dynamics. In Malang Regency, these fluctuations impact inflation and economic stability, necessitating an accurate forecasting model to support decision-making. This study aims to develop a price forecasting model using Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA) methods and compare their performance using daily time series data on large red chili prices from January 2022 to August 2024, obtained from the Representative Office of Bank Indonesia in Malang. The data underwent preprocessing, where LSTM data was transformed using MinMaxScaler, while ARIMA data was differenced to meet stationarity assumptions, then split into 80% training and 20% testing data, with optimal parameters obtained through Grid Search for both models. The results show that the LSTM model with three layers (150, 150, 150 units) and a dropout of 0.2 achieved an RMSE of 2.326 and MAPE of 3.65%, whereas the best ARIMA configuration (4,1,3) achieved an RMSE of 2.455 and MAPE of 3.80%. Although both models performed competitively and yielded promising results, LSTM demonstrated superior accuracy in forecasting large red chili prices in dynamic market conditions
Market Basket Analysis untuk Penjualan Perlengkapan Cetak dengan Algoritma FP Growth
Full text linkPerusahaan yang bertumbuh adalah perusahaan yang terus berkembang dan berinovasi menemukan berbagai macam strategi seiring dengan berjalannya waktu agar meningkatkan omzet usaha yang ditandai dengan penjualan barang. Namun apabila perusahaan serupa atau kompetitor juga melakukan pendekatan strategi yang sama, maka perlu mempersiapkan strategi pemasaran baru untuk meningkatkan penjualan.
Market Basket Analysis merupakan pendekatan analisis data untuk mengenali pola perilaku konsumen terhadap keterkaitan antar produk dalam transaksi penjualan. Metode yang digunakan dalam analisis ini adalah association rule mining, yang berfokus pada pencarian relasi produk yang dibeli secara bersamaan. Terdapat tiga metrik utama dalam metode ini, yaitu support, confidence, dan lift, yang digunakan untuk menilai relevansi aturan asosiasi. Algoritma FP-Growth dipakai karena mampu menemukan aturan asosiasi secara lebih efisien melalui pembuatan struktur data FP-Tree, yang memungkinkan penemuan frequent itemset tanpa perlu menghasilkan kombinasi kandidat secara eksplisit.
Pengujian dilakukan pada data transaksi penjualan dari tahun 2022-2023 dengan total sebanyak 118.709 transaksi dengan bahasa Python lalu menghasilkan 9 aturan asosiasi. Pelaku bisnis dapat melakukan strategi pemasaran seperti membuat promosi product bundling dan peletakan produk yang berdekatan. Produk-produk tertentu yang memiliki keterkaitan satu sama lain seperti HEAD L210 L1110 L3110 L3150 DUS KECIL NEW dengan FP HEAD CLEANER PREMIUM 20ML (93,99%) dan FP PERMANENT STAMP 10ML – BLACK dengan FP PERMANENT STAMP REMOVER 5ML (97,91%) dapat menjadi kandidat bundel produk yang menjanjikan dikarenakan memiliki nilai confidence yang tinggi
Pengembangan Sistem Informasi Estimasi Biaya Proyek Perangkat Lunak Berbasis Function Point
Full text linkPenelitian ini bertujuan untuk mengembangkan sistem informasi yang dapat mengotomatiskan proses estimasi biaya proyek perangkat lunak secara akurat dan efisien. Sistem ini mengadopsi metode Function point Analysis untuk menentukan ukuran fungsional perangkat lunak, kemudian menghitung estimasi biaya berdasarkan kompleksitas setiap komponen yang dinilai oleh manajer proyek. Pengembangan sistem menggunakan bahasa pemrograman VB.NET dengan arsitektur client-server dan database Microsoft SQL Server. Metode waterfall digunakan dalam tahapan pengembangan, memastikan setiap fase, mulai dari analisis kebutuhan hingga pengujian, dilaksanakan secara sistematis. Hasil penelitian menunjukkan bahwa sistem yang dikembangkan mampu menghasilkan estimasi biaya yang lebih akurat dan cepat dibandingkan metode konvensional. Sistem ini juga menyediakan fitur pelaporan komprehensif yang memudahkan manajemen proyek dalam mengambil keputusan yang tepat. Fitur ini meliputi laporan rinci mengenai komponen proyek, estimasi waktu, dan sumber daya yang dibutuhkan. Selain itu, sistem ini mampu menghasilkan harga proyek yang dapat digunakan sebagai dasar penentuan harga penawaran kepada klien, meningkatkan transparansi dan kepercayaan dalam negosiasi proyek. Sistem informasi yang dikembangkan tidak hanya membantu dalam mengelola biaya proyek secara lebih efisien, tetapi juga meningkatkan efisiensi operasional perusahaan secara keseluruhan. Dengan menyediakan estimasi yang akurat dan laporan yang komprehensif, manajemen proyek dapat merencanakan dan mengalokasikan sumber daya dengan lebih efektif. Kesimpulannya, sistem informasi ini dapat menjadi alat bantu yang sangat berguna bagi perusahaan dalam mengelola proyek perangkat lunak, meningkatkan akurasi estimasi biaya, dan mempercepat proses pengambilan keputusan, sehingga perusahaan dapat bersaing lebih baik di pasar yang kompetitif
Comparative Analysis of Neural Network Architecture Optimization: A Study on Genetic Algorithm, Random Search, Grid Search, and Adaptive Search Methods for Digit Classification
Full text linkThis research presents a comprehensive comparative analysis of four neural network architecture optimization methods: Genetic Algorithm (GA), Random Search, Grid Search, and Adaptive Search. Using the MNIST digits dataset, a systematic evaluation was performed based on accuracy, computational efficiency, and architectural complexity. The experimental results demonstrate that the Genetic Algorithm achieved the highest accuracy at 98.33%, while Grid Search demonstrated computational efficiency with the fastest execution time at just 31.06 seconds. Random Search and Adaptive Search showed competitive performance with accuracies of 97.78% and 97.22% respectively, with varying computational requirements. The study revealed that simpler architectures with one or two layers often performed comparably to more complex structures, challenging the common assumption that deeper networks necessarily yield better results. The Genetic Algorithm converged to an optimal single-layer architecture with 119 neurons and ReLU activation, while Adaptive Search explored a more complex three-layer solution. The research identified a non-linear relationship between accuracy gains and computational costs, indicating that substantial increases in computational investment may yield diminishing returns in performance improvement. The convergence patterns of each method provided additional insights, with GA showing steady improvement across generations while Random Search achieved early discovery of good solutions. These findings contribute to both theoretical understanding and practical applications of neural network optimization, offering valuable insights into the trade-offs between methods and practical guidelines for selecting appropriate architecture optimization strategies based on specific requirements for accuracy and computational constraints
Faktor Pengaruh Peralihan ke Online Learning pada Pegawai Negeri Sipil Berbasis Teori Push Pull Mooring
Full text linkPegawai Negeri Sipil (PNS) diwajibkan untuk terus mengembangkan kompetensi minimal 20 jam per tahun, yang umumnya dilakukan melalui pembelajaran tradisional Pandemi COVID-19 tahun 2020 mengubah paradigma ini dan menjadikan Online Learning sebagai solusi utama akibat pembatasan sosial dan lockdown untuk mengendalikan penyebaran virus. Bahkan, dalam perkembangannya, banyak instansi pemerintah mengadopsi konsep Corporate University untuk mendukung Online Learning, meskipun banyak menghadapi tantangan, seperti wilayah yang luas dan tersebar pada 38 provinsi dan 514 Kabupaten/Kota dan literasi digital yang rendah. Namun, seiring dengan meredanya pandemi, Online Learning menjadi suatu pilihan, bukan lagi keharusan dan pembelajaran dengan Traditional Learning kembali dibuka. Penelitian ini berusaha mengungkap faktor-faktor Pendorong (push) yang memengaruhi PNS untuk meninggalkan Traditional Learning, faktor-faktor yang menjadi Penarik (Pull) yang memengaruhi PNS untuk beralih pada Online Learning, dan faktor Penambat (Mooring) yang memengaruhi PNS untuk tetap menggunakan Traditional Learning atau beralih pada Online Learning. Berdasarkan data dari 463 responden PNS yang pernah menggunakan Traditional Learning maupun Online Learning yang diolah dengan metode analisa Structural Equation Model (SEM) dengan bantuan aplikasi SPSS dan AMOS, dapat diketahui bahwa seluruh variabel dalam Pull Factor secara keseluruhan memengaruhi keinginan berpindah, dan tidak seluruh variabel dalam Push Factor dan Mooring Factor yang memengaruhi keinginan berpindah
A Hybrid Machine Learning and Deep Learning Approach for In-Game Assistance
Full text linkThe rapid development of artificial intelligence (AI) has opened new possibilities for enhancing user interaction within video games. This study presents the design and implementation of a button-based assistant system for the simulation game Story of Seasons: Friends of Mineral Town, aimed at simplifying repetitive player tasks and improving the overall gameplay experience. The proposed system leverages a hybrid approach that combines Machine Learning and Deep Learning techniques, specifically Optical Character Recognition (OCR) with Tesseract, object detection using a custom-trained YOLOv7 model, the A* pathfinding algorithm for navigation, and automated input control through scripting. The assistant is capable of reading in-game time, weather, and events directly from screen captures, recognizing non-player characters (NPCs), and automatically directing the player’s character to desired locations or NPCs based on contextual data such as day, time, and weather conditions. A database-driven module stores key information such as NPC schedules, favorite gifts, and daily events to enable informed decision-making and interaction automation. Comprehensive testing was conducted, including comparisons of pathfinding algorithms, model accuracy assessments, and user experience evaluations involving volunteers. Results showed high detection accuracy with YOLOv7 and positive user feedback on the assistant's interface and usability. Users reported a more streamlined and enjoyable gaming experience, especially in managing daily tasks and character interactions. This research demonstrates how a hybrid AI-based approach can be effectively applied to traditional video games, offering a foundation for future development in intelligent game assistance systems. The proposed methodology not only improves convenience but also provides insights into the practical integration of AI in user-centric game design
Multi View Natural Network for Cross-Project Software Defect Prediction
Full text linkSoftware Defect Prediction (SDP) plays a critical role in software engineering by enabling early identification of potentially defective modules, to assist developers and testers in prioritizing testing and inspection efforts to improve software quality and reliability. Driven by rapidly changing business requirements, defect prediction models have become increasingly essential in quality assurance workflows. Traditional approaches to SDP focused on Within-Project Defect Prediction (WPDP), where models are trained on historical data from the same project and effective under sufficient data conditions. This challenge motivates the adoption of Cross-Project Defect Prediction (CPDP), which leverages data from different projects. However, CPDP faces notable challenges including datasets distributional differences and class imbalance, which can degrade prediction performance and bias. To address these issues, recent studies have proposed data transformation, resampling, and domain adaptation techniques. In this study, we explore a multi-view learning approach using Neural Networks (NN) to enhance generalization and performance in CPDP scenarios. By leveraging multiple views of the same dataset—generated through concatenation of heterogeneous software metrics, imputation for missing values, normalization using Box-Cox transformation, and embedding-based feature transformation—we aim to construct a robust Multi-View Neural Network (MVNN). This architecture enables the integration of diverse information while mitigating the limitations of single-view learning in CPDP. Our method preserves more in-formation compared to conventional approaches that rely only on shared features. Experimental validation using benchmark SDP repositories demonstrates the competitiveness of our approach, offering improved performance over existing CPDP models and highlighting the potential of multi-view learning in defect prediction tasks
Classification of Skin Diseases Using Transfer Learning with ResNet-50 Architecture and Data Preprocessing Using Real-ESRGAN and Wiener Filter
Full text linkThe skin is a vital organ which serves as a barrier against external factors, yet it’s highly susceptible to diseases. These diseases are often presented as lesions with similar appearances, making it difficult to be diagnosed and prone to human errors. To address this challenge, this study uses Deep Learning, particularly the ResNet-50 architecture using Transfer Learning, to classify skin diseases from lesion. In this study, data augmentation is implemented to increase dataset size, thus improving model performance and preventing overfitting. Data is then preprocessed using Real-ESRGAN to enhance resolution and the Wiener Filter to sharpen the features. Adam optimizer is used to further enhance the model’s performance. Hyperparameter tuning is also implemented to optimize the model parameters, and dropout regularization is applied to enhance the model's ability to be able to accurately classify unseen data. The model managed to achieve a high accuracy of 99.09%, with 0.96 precision, 0.95 recall, and 0.95 F1-score. These results demonstrate the effectiveness of combining Real-ESRGAN and Wiener Filter with the ResNet-50 architecture and the Adam optimizer in developing a robust model for skin disease classification. This approach offers a promising tool for healthcare professionals which may help reduce human error in dermatological diagnosis