Publication Search

Studi ini dirancang untuk menginvestigasi pengaruh Suhu Permukaan Laut (SST) dan konsentrasi klorofil-a terhadap variabilitas hasil tangkapan ikan di kawasan Pantai Selatan melalui pendekatan model regresi linear berganda dengan estimasi Ordinary Least Squares (OLS). Pada tahap diagnostik awal, terdeteksi adanya indikasi autokorelasi positif pada residual model (statistik Durbin–Watson = 0,993), suatu kondisi yang berpotensi menginduksi bias pada estimasi standard error sehingga dapat mengkompromikan validitas inferensi statistik. Guna mengantisipasi permasalahan tersebut, dilakukan koreksi menggunakan metode Newey-West Heteroskedasticity and Autocorrelation Consistent (HAC) standard errors yang diimplementasikan melalui komputasi Python. Hasil estimasi pasca-koreksi mengungkapkan bahwa secara simultan, variabel SST dan klorofil-a memberikan pengaruh yang signifikan terhadap keragaman hasil tangkapan ikan, dengan nilai koefisien determinasi (R²) sebesar 23,7%. Berdasarkan uji parsial (uji t terkoreksi), SST menunjukkan pengaruh negatif yang signifikan secara statistik terhadap hasil tangkapan, sementara klorofil-a tidak memperlihatkan pengaruh yang signifikan. Temuan tersebut mengimplikasikan bahwa variabel termal (SST) memiliki kontribusi yang lebih dominan dibandingkan indikator produktivitas primer (klorofil-a) dalam memengaruhi dinamika hasil tangkapan ikan pada wilayah kajian.

Ketidakseimbangan antara tingkat persediaan dan permintaan pasar dapat mengakibatkan kelebihan stok, meningkatnya biaya penyimpanan, dan meningkatnya risiko kerusakan produk. PT. Tridaya Sakti Medima, sebuah perusahaan distribusi farmasi yang berlokasi di Palembang, menghadapi tantangan serupa dalam mengelola persediaannya. Penelitian ini bertujuan untuk menganalisis pola pembelian dengan menerapkan teknik penambangan data menggunakan algoritma Apriori dan untuk merancang sistem rekomendasi produk yang mendukung pengambilan keputusan dalam pengendalian persediaan. Studi ini mengadopsi metodologi CRISP-DM (Cross Industry Standard Process for Data Mining), yang terdiri dari tahap pemahaman bisnis, persiapan data, pemodelan, evaluasi, dan penerapan. Data transaksional yang dianalisis terdiri dari catatan penjualan historis dari Januari hingga Desember 2024, dengan total 5.410 entri setelah proses pembersihan data. Analisis dilakukan menggunakan Python, sedangkan Streamlit digunakan untuk mengembangkan dasbor interaktif untuk visualisasi. Temuan menunjukkan bahwa algoritma Apriori berhasil mengidentifikasi aturan asosiasi antarproduk berdasarkan metrik dukungan dan kepercayaan. Nilai kepercayaan dan peningkatan tertinggi ditemukan pada kombinasi produk (ITR) PECTORIN SYRUP 120 ML @ 24 dan (ITR) ITRABAT SYR 100 ML @ 24 dengan rasio lift 16,43. Hasil ini disajikan melalui dasbor yang dirancang untuk membantu PT. Tridaya Sakti Medima dalam meningkatkan manajemen persediaan dan mengatasi masalah ketidakseimbangan stok.

Perkembangan media sosial yang pesat telah menghasilkan volume data teks yang sangat besar dalam bentuk opini dan ulasan pengguna. Analisis manual terhadap data ini untuk memahami sentimen publik tidak lagi efisien. Hal inilah yang mendasari dibuatnya sebuah aplikasi analisis sentimen otomatis menggunakan metode machine learning Support Vector Machine (SVM). Untuk mendukung penelitian yang dilakukan, peneliti menggunakan metode penelitian Research and Development (R&D) serta perancangannya menggunakan metode Object Oriented Programming (OOP) menggunakan Unified Modeling Language (UML). Aplikasi ini dibangun menggunakan bahasa pemrograman Python dengan framework Streamlit untuk menyediakan antarmuka pengguna yang interaktif dan mudah diakses. Aplikasi yang dihasilkan berhasil mengotomatiskan seluruh proses, memungkinkan pengguna untuk memasukkan teks baru dan mendapatkan hasil analisis sentimen secara real-time. Penelitian ini menunjukkan bahwa kombinasi SVM dan Streamlit dapat menjadi solusi yang efektif dan praktis untuk analisis sentimen otomatis.

This study applies the Mean Variance model, which aims to form an optimal portfolio composition in the health, property, and cyclical consumer sectors and combine the three sectors into one portfolio, then visualize its efficient frontier. This study analyzes the return profiles and compares the risks of each portfolio using alternative risk measures such as the Coefficient of Variation (CV), Value at Risk (VaR), and Conditional Value at Risk (CVaR). Daily closing price data for the three sectors listed on the Indonesia Stock Exchange (IDX) from March 2, 2020, to March 3, 2025, were used in this study. Stock selection was conducted using purposive sampling, followed by selecting seven stocks for optimization based on the lowest Coefficient of Variation (CV) value. Portfolio optimization analysis was conducted using the Python programming language with Visual Studio Code software. The findings of this study indicate that the combined portfolio incorporating the three sectors is the most efficient, with an expected return of 0.104%, standard deviation of 0.007, and alternative risk measures such as Coefficient of Variation (CV) 6.9328, Value at Risk (VaR) of -0.99%, and Conditional Value at Risk (CVaR) of -1.44%, which are lower than those of single-sector portfolios. Visualization of the efficient frontier curve confirms that the combined portfolio offers better results in terms of risk and return. The results of this study indicate that cross-sector diversification can significantly reduce risk and prevent significant losses.

Polusi udara partikulat halus (PM2,5) merupakan ancaman serius bagi kesehatan masyarakat di Kabupaten Brebes, Jawa Tengah. Faktor penyumbang utamanya adalah emisi kendaraan di jalur Pantura, aktivitas industri perikanan, serta konsentrasi tinggi selama musim kemarau (Juni–November). Tidak adanya model peramalan sub-jam yang akurat menghambat pengembangan sistem peringatan dini yang efektif. Penelitian ini mengembangkan dan mengevaluasi model deep learning berbasis Transformer untuk memprediksi konsentrasi PM2,5 dengan resolusi waktu 15 menit. Data yang digunakan berasal dari NASA GEOS-CF (band PM25_RH35_GCC) yang diakses melalui Google Earth Engine menggunakan API Python. Dataset mencakup periode 1 Januari hingga 22 November 2025, menghasilkan 7.813 observasi per jam, yang kemudian diinterpolasi linear menjadi 31.249 titik data dengan resolusi 15 menit. Arsitektur Transformer terdiri dari 3 lapis enkoder, 4 kepala perhatian multi-head, dimensi embedding 128, dimensi feed-forward 256, panjang sekuen 60 timestep, dan augmentasi fitur menggunakan rerata bergulir (*rolling mean*, jendela = 3) dan beda pertama (*first difference*). Pelatihan dilakukan dengan TensorFlow-Keras, pengoptimal Adam, penjadwal peluruhan kosinus (*cosine decay scheduler*), dan fungsi kerugian Huber. Pembagian data dilakukan secara kronologis: 70% pelatihan, 30% validasi. Evaluasi pada set uji independen (16 Agustus–21 November 2025, 9.357 observasi atau 97 hari 11 jam 15 menit) menghasilkan MAE 0,7691 µg/m³, RMSE 1,2052 µg/m³, R² 0,9945, dan *Explained Variance Score* 0,9948. Model ini mampu menggambarkan variasi diurnal dan anomali musiman secara akurat, jauh melampaui model LSTM dan GTWR konvensional. Penelitian ini memberikan kontribusi signifikan di bidang Teknologi Informasi melalui kerangka kerja pengolahan *big data* satelit untuk aplikasi lingkungan.

Hijaiyah letters have varying shapes, and some of them are very similar, often causing errors in the manual character recognition process. This study aims to classify Hijaiyah letters based on digital images using the Convolutional Neural Network (CNN) method. This method was used in this study with a dataset consisting of 28 letter classes and a total of 4,480 images obtained from various public sources and private data. All images underwent a preprocessing stage that included labeling, resizing, normalization, and augmentation, then were divided into three parts, namely training data, validation data, and test data with a ratio of 70:20:10. The training process was carried out using the Python programming language with the help of the TensorFlow and Keras libraries on the Google Colab platform. The test results showed that the CNN model achieved an accuracy of 97.10%, with an average precision, recall, and F1-score of 0.97, respectively. Classification errors only occurred in letters that had similar shapes, such as Syin and Sin. Based on these results, the CNN method proved to be effective, efficient, and accurate in recognizing Hijaiyah letter image patterns, so it can be used as a basis for developing classification models with higher accuracy in the future.  

Penelitian ini bertujuan untuk membuat sistem yang bisa memprediksi jumlah mahasiswa menggunakan pendekatan machine learning dengan Python dan antarmuka Streamlit sebagai alat interaktif. Masalah utama yang dibahas adalah kebutuhan lembaga pendidikan tinggi dalam meramalkan jumlah mahasiswa di masa depan agar bisa membantu dalam merencanakan kapasitas dan mengalokasikan sumber daya secara tepat. Solusi yang ditawarkan adalah dengan menggabungkan dua metode prediksi, yaitu Regresi Linear dan ARIMA. Regresi Linear digunakan untuk mencari pola dasar dalam data masa lalu, sedangkan ARIMA dipakai untuk menganalisis data yang memiliki perubahan atau ketergantungan terhadap data sebelumnya. Data yang digunakan berasal dari Pangkalan Data Pendidikan Tinggi (PDDIKTI) untuk program studi Ilmu Komputer di Universitas Negeri Medan pada periode tahun 2019 sampai 2024. Sistem ini dirancang agar orang yang tidak ahli di bidang teknologi bisa melakukan analisis secara visual melalui grafik dan metrik evaluasi yang ditampilkan langsung di aplikasi Streamlit. Hasil pengujian menunjukkan bahwa Regresi Linear memberikan akurasi tinggi pada data yang memiliki tren stabil dengan nilai MAPE di bawah 10%, sedangkan ARIMA memberikan prediksi yang lebih baik pada data yang tidak stabil. Penerapan sistem ini menunjukkan bahwa menggabungkan metode statistik dan pembelajaran mesin bisa meningkatkan efisiensi dan ketepatan dalam analisis data akademik. Temuan ini memberikan manfaat nyata dalam membantu pengambilan keputusan strategis terkait perencanaan jumlah mahasiswa, serta membuka kemungkinan pengembangan model yang lebih canggih di masa depan.

Pengelolaan persediaan pada bisnis penjualan suku cadang motor sering menghadapi tantangan ketidakseimbangan stok akibat kurangnya pemahaman terhadap pola pembelian konsumen. CV FOLC MORA, sebagai salah satu pengecer sparepart di Palembang, mengalami ketidakefisienan dalam pengendalian stok yang dapat mengakibatkan biaya penyimpanan berlebih atau hilangnya peluang penjualan ketika barang populer tidak tersedia. Kebaruan dari penelitian ini terletak pada konteks penerapan algoritma Apriori yang difokuskan pada skala usaha mikro, kecil, dan menengah (UMKM) penjualan suku cadang motor, yang relatif jarang dieksplorasi pada penelitian terdahulu yang umumnya menitikberatkan pada e-commerce berskala besar. Selain itu, penelitian ini tidak hanya menghasilkan analisis pola pembelian, tetapi juga mengintegrasikan hasil tersebut ke dalam dashboard interaktif berbasis Streamlit yang memudahkan pelaku UMKM dalam melakukan pemantauan stok dan perencanaan pembelian. Kombinasi konteks UMKM dan pemanfaatan dashboard analitik inilah yang menjadi nilai kebaruan sekaligus kontribusi praktis penelitian ini. Penelitian ini bertujuan untuk menganalisis pola pembelian menggunakan algoritma Apriori guna mendukung pengambilan keputusan dalam optimalisasi stok dan peningkatan penjualan. Metode yang digunakan adalah CRISP-DM (Cross-Industry Standard Process for Data Mining) yang terdiri dari enam tahapan: pemahaman bisnis, pemahaman data, persiapan data, pemodelan, evaluasi, dan implementasi. Data transaksi penjualan yang dianalisis berasal dari periode Januari hingga April 2025, dengan total 1.550 transaksi dan 6.478 data item. Pengolahan data dilakukan menggunakan Python dengan bantuan pustaka Pandas dan MLxtend, sedangkan hasil analisis disajikan melalui dashboard interaktif berbasis Streamlit. Algoritma Apriori menghasilkan keterkaitan kuat antar produk, seperti “Ban Luar Swallow 80/90-17” dan “Velg TK Excel Rim” dengan nilai support 1%, confidence 76%, dan lift 2,15. Temuan ini menjadi dasar penyusunan strategi bundling produk dan keputusan restok barang. Dashboard yang dikembangkan memudahkan pemilik usaha untuk mengunggah data transaksi, mengatur parameter analisis, serta melihat frequent itemset dan rekomendasi kombinasi produk. Pendekatan ini meningkatkan efisiensi pengelolaan stok, mendukung strategi pemasaran berbasis data, serta berkontribusi pada peningkatan kepuasan pelanggan dan kinerja penjualan.

Handwritten digit recognition is one of the key challenges in the field of digital image processing and artificial intelligence, with significant potential in various applications such as automatic form input systems, handwritten data correction, and attendance systems based on handwriting. This study aims to develop a web-based information system capable of automatically recognizing handwritten digits using the K-Nearest Neighbors (KNN) classification method. The system is designed through several main stages, including image preprocessing (conversion to grayscale, thresholding, and image size normalization), feature extraction using the zoning technique, and classification using the KNN algorithm. This research utilizes the MNIST dataset, which contains thousands of handwritten digit images ranging from 0 to 9. The system is developed using the Google Colab platform, supported by Python libraries such as OpenCV, NumPy, and Scikit-learn. Test results show that the system can achieve an accuracy of over 90% at certain K values, indicating that the KNN method is quite effective and efficient in recognizing handwritten digit patterns. This system is expected to be applicable for various digitization needs of handwritten numbers in education, administration, and information technology sectors

This training program is designed to strengthen digital literacy and basic skills in data analysis for high school students in the Saboking-king area, Palembang, by introducing the Python programming language and the use of Google Colab as a cloud-based programming platform. The approach applied is project-based learning and hands-on experience, which includes planning, implementation, and evaluation stages. Based on the implementation results, the majority of participants showed a good understanding of Python basics and successfully completed the mini-project assignments. The final evaluation revealed an increase in students' motivation and interest to explore programming further. The success of this program confirms that the learning methods and media used are able to support students' understanding of the basic concepts of data analysis. Similar activities are recommended to be implemented in other areas that have limited access to technology.

Customer satisfaction is a crucial indicator in assessing the quality of a company's products, services and overall experience. This research aims to identify the level of customer satisfaction and optimize the available data for effective use in sentiment analysis. In this study, we analyzed 4,353 customer reviews collected over the past year, with 3,481 reviews used as training data and 871 reviews as testing data. The analysis process was conducted using the Cross-Industry Standard Process for Data Mining (CRISP-DM) approach and leveraged the Logistic Regression algorithm to build a predictive model. Model evaluation using the confusion matrix yielded an accuracy of 94.60%, a precision of 94.26%, and a recall of 94.60%. The analysis was conducted using Jupyter Notebook and the Python programming language. The results indicate that sentiment analysis is effective in identifying and predicting customer satisfaction levels, which in turn can help a company’s products improve its service strategies. The optimization of previously underutilized data now provides deeper insights into customer perceptions and expectations, enabling the company to make more targeted decisions and enhance overall customer satisfaction.

Face detection and facial landmarks are an important technique in the field of computer vision with a wide range of potential applications, including expression recognition, security systems, and human-computer interaction. This study explores the implementation of facial landmarks detection using Python and OpenCV, focusing on the use of the Haar Cascade algorithm for face detection and the Local Binary Features (LBF) model for the identification of landmarks. The proposed method implements real-time detection via webcam, capable of recognizing 68 important points on the human face. The results show that the approach using OpenCV and LBF models has good accuracy in detecting and tracking facial features in different lighting conditions and viewing angles. This research contributes to the development of efficient and reliable facial detection methods, with wide application potential in the fields of computer vision, security, and behavioral analysis.

Customer segmentation plays a pivotal role in driving marketing strategies and improving customer retention across various industries. This study explores the application of the RFM (Recency, Frequency, Monetary) model for customer segmentation in a Software-as-a-Service (SaaS) business, using Python for efficient data processing and analysis. By analyzing one year of customer purchase data, we segmented customers into key groups such as "Champions," "Loyal Customers," and "At Risk." The results highlight that targeted discount strategies significantly affect profitability, especially for high-value customer segments. Furthermore, the research builds upon existing methodologies, demonstrating how Python-based implementations streamline RFM analysis and allow for scalable solutions in business contexts, as illustrated in prior works by Hermawan et al. (2024). This study offers actionable recommendations, including tailored discounting, loyalty programs, and personalized engagement strategies, to enhance customer retention and business profitability. The findings underscore the importance of data-driven marketing approaches for customer segmentation and engagement, reinforcing the relevance of the RFM model in modern business environments.

This study aims to optimize marketing strategies through RFM (Recency, Frequency, Monetary) analysis on a retail transaction dataset obtained from Kaggle. The dataset contains 64,682 transactions from 5,242 SKUs involving 22,625 customers over one year. Data cleaning and RFM analysis were conducted to segment customers based on recency, frequency, and monetary values. The findings reveal that customers were segmented into groups such as Champions, Loyal Customers, and At Risk. These segments provide valuable insights for developing targeted marketing strategies, such as loyalty programs for high-value customers and retention campaigns for at-risk customers. The study demonstrates that RFM analysis is effective in identifying valuable customer segments and optimizing marketing efforts based on customer behavior. This approach can increase customer retention and improve the return on investment (ROI) in marketing campaigns.

Pada tahun 2024, Indonesia akan menyelenggarakan pemilihan umum serentak yang meliputi pemilihan presiden dan pemilihan wakil rakyat di seluruh Indonesia. Masyarakat menanggapi kejadian ini dengan perasaan campur aduk, membagikan pemikirannya di situs media sosial seperti Twitter. Penelitian analisis sentimen calon presiden Indonesia tahun 2024 dilakukan terkait peristiwa ini. Sebanyak 1458 tweet digunakan dalam penelitian ini. Dengan 40,31% responden menyatakan sikap positif dan 43,46% menyatakan sentimen negatif, temuan analisis menunjukkan keseimbangan antara kedua sentimen tersebut. Menggunakan frasa "calon presiden," program Python di situs web Google Colab mengambil data twitter. Pendekatan K-Nearest Neighbor digunakan dalam proses klasifikasi. Selain itu data latih dibagi 6 : 4. 40% data uji dan 60% data latih. Nilai evaluasi yang diperoleh dari pengujian model dengan teknik K-Nearest Neighbor adalah akurasi sebesar 90,95%, presisi sebesar 62,17%, recall sebesar 62,33%, dan F-Measure sebesar 61,87%.

In a competitive business environment, understanding customer behaviour and improving retention strategies are critical to a company's success. Many companies struggle to identify valuable customers, understand their needs, and develop effective marketing strategies. One method that has proven effective is Recency, Frequency, and Monetary (RFM) analysis, which measures customer value based on three dimensions: when the customer last made a purchase, how often they transact, and how much money they spend. This research focuses on applying the RFM method with Python for customer segmentation in a retail company. By analysing customer transaction data, this research shows how RFM analysis can provide deep insights into customer behaviour and assist in the development of more targeted marketing strategies. The ultimate goal is to improve customer retention and maximise the return on investment (ROI) of marketing activities. This research offers practical solutions to common challenges in customer relationship management and contributes to the development of more efficient data-driven marketing methods.

MSIB (Certified Independent Study and Internship) is one of the activity programs at the Merdeka Campus which aims to help students improve their skills and develop themselves. MSIB appointed Orbit Future Academy as one of the partners in the Independent Study program. Founded in 2016 with the aim of improving the quality of life through innovation, education and skills training. In accordance with its mission, namely "We curate and localize international programs and courses for upskilling, re-skilling youth, and the workforce towards jobs of the future". Partners provide opportunities for students to take Artificial Intelligence programs and study online. Learning consists of eight material courses including Python Programming, AI Technology Logic and Concepts, AI Project Cycle, AI Research Methods, ChatGPT, Professional and Company Ethics, Financial Literacy and ending with a Final Project. The final project scope carried out is the Occupancy Detection in Parking Lot project. This project uses the Computer Vision domain with the selection of the YOLO model in detecting objects and pixel segmentation. The project begins with selecting a dataset using roboflow which then goes through data pre-processing for cloning, annotation and augmentation. Then the model is trained using machine learning and deep learning algorithms to understand patterns and characteristics related to parking spaces. Once trained, the AI model will be validated using test data. This aims to ensure that the model truly recognizes the presence of the vehicle. Next, form the application design in creating an informative interface using wireframes. Then enter the deployment stage so that the system can be accessed widely and easily via the web. Lastly, field testing is to find out the performance of the application that has been designed.