METHOMIKA: Jurnal Manajemen Informatika & Komputerisasi Akuntansi

Optimalisasi Prediksi Klaim Kendaraan Bermotor Menggunakan Time Series Analysis dengan Pendekatan Variasi Kalender dalam Perusahaan Asuransi Umum

2026-03-15T11:37:57+07:00

General insurance companies face significant challenges in predicting motor vehicle insurance claims due to seasonal fluctuations and calendar variations. This study aims to optimize the accuracy of claim predictions using the Autoregressive Integrated Moving Average (ARIMA) method combined with a calendar variation approach. Data analysed includes nominal claim values from January 2015 to December 2023. Regression analysis results show that calendar variations, specifically periods "before" and "during" the Eid al-Fitr holiday, significantly impact claim reductions. The best-fit time series model obtained is ARIMA (1,0,1)(0,1,1)₄. This model satisfies white noise and normality assumptions for residuals, providing a robust framework for data-driven decision-making, reserve allocation, and operational readines.

Perancangan Sistem Peringatan Dini Bencana Berbasis Website Terintegrasi BMKG dan Notifikasi Whatsapp

2026-03-15T09:13:59+07:00

Indonesia has a high level of disaster risk due to its geological conditions. Although the BMKG has provided early warning information, its dissemination has not been fully received by the public in a timely manner. Meanwhile, the widespread use of WhatsApp in Indonesia offers an opportunity to utilize it as a medium for delivering disaster-related information. This study aims to design a website-based disaster early warning system integrated with BMKG data and WhatsApp notifications based on provincial regions. The system was developed using the Waterfall method with PHP, MySQL, and a WhatsApp gateway service, utilizing earthquake and weather data that are updated automatically. The results indicate that the system operates according to functional requirements. WhatsApp notifications were successfully delivered with a 95% success rate and an average delivery time of 2.6 seconds. Black box testing showed 100% valid results, while the System Usability Scale evaluation achieved a score of 82.5, indicating a very good level of usability.

Media Pembelajaran Interaktif Berbasis Augmented Reality Pengenalan Anggota Tubuh dalam Bahasa Kutai untuk Sekolah Dasar Negeri 008 Samarinda Utara

2026-03-14T09:26:54+07:00

This study develops markerless Augmented Reality (AR) interactive learning media to introduce human body parts in the Kutai language as local content (Mulok) at SDN 008 Samarinda Utara. The primary issue stems from traditional methods—textbooks, 2D images, and conventional lectures—causing student boredom, low motivation, and shallow vocabulary mastery. This research aims to enhance learning motivation, conceptual understanding, and Kutai language proficiency through immersive experiences. The development follows the six-stage Multimedia Development Life Cycle (MDLC). The application was built on the Assemblr EDU platform (browser-based WebAR) using 3D models from Blender, interactive buttons, and Kutai language explanations. Black box testing was successful, while beta testing with ten students yielded an average score of 76.9% (Good category). This media supports immersive learning without additional installations, facilitating effective Mulok implementation at SDN 008. Future research is recommended to expand the respondent pool and integrate advanced interactive features, such as AR-based quizzes, audio narration, and animations, to further enhance overall media effectiveness. Exploring alternative platforms, including Unity and Zappar, is also encouraged. This tool represents a significant advancement in digitalizing regional language education, ensuring that local indigenous knowledge remains engaging and accessible for the younger generation in East Kalimantan.

Implementasi Dynamic Rendering untuk Optimasi Efisiensi Server dan Indeksabilitas Pada Single Page Application

2026-03-23T00:35:55+07:00

The development of a Client Side Rendering (CSR)-based Single Page Application (SPA) has search engine indexability weaknesses, while Server Side Rendering (SSR) overloads server computation. This study aims to prove the efficiency of the Dynamic Rendering architecture as a middle-ground solution. Through Nginx configuration, the server detects the User-Agent to serve static pages (CSR) to human users and fully rendered pages (SSR) to bot crawlers. Experimental testing was conducted on a Virtual Private Server (VPS) using k6 with a constant load of 100 Virtual Users, Prometheus, and Grafana, as well as Document Object Model (DOM) validation via Google Search Console. The Mann-Whitney U Test results proved a significant performance difference with a p-value < 0.05. The implementation of Dynamic Rendering is highly efficient, capable of reducing CPU Utilization by 92.28% and Memory Usage by 5.14%, and increasing Request Per Second (RPS) capacity by 17.19%. Indexability validation also confirmed that crawlers successfully received the HTML document entirely. In conclusion, Dynamic Rendering is proven to be an effective architectural solution to minimize server load while ensuring optimal content visibility on search engines.

Klasifikasi Adopsi Berbasis Kecerdasan Buatan pada UMKM di Indonesia Menggunakan Algoritma Random Forest

2026-03-23T00:28:13+07:00

Micro, Small, and Medium Enterprises (MSMEs) play a strategic role in the Indonesian economy; however, digital transformation based on artificial intelligence (AI) remains a significant challenge. This study aims to classify AI adoption among MSMEs in Indonesia using the Random Forest algorithm and to identify the factors that influence it. The dataset was obtained from the Zenodo repository, consisting of questionnaire results regarding AI adoption in MSMEs. The research stages included data cleaning, encoding, splitting the data into training (80%) and testing (20%) sets, implementing the Random Forest algorithm, evaluation, and result analysis. The evaluation results show an accuracy of 80.3% with an ROC-AUC of 0.884. The weighted precision, recall, and F1-score values are 81.2%, 80.3%, and 80.4%, respectively. These evaluation results indicate that the Random Forest algorithm performs well on this dataset. Furthermore, the feature importance analysis revealed several influential variables in AI adoption among MSMEs, including strategic decision-making (10.9%), digital leadership (8.3%), and respondent position (7.8%). In conclusion, the implementation of the Random Forest algorithm demonstrates strong performance in classifying AI adoption among MSMEs in Indonesia and highlights key influential variables such as strategic decision-making, digital leadership, and respondent position.

Simulasi Reaksi Kimia Berbasis Algoritma Titrasi Menggunakan Python

2026-03-08T08:49:52+07:00

Learning acid–base titration concepts in schools is often constrained by limited laboratory facilities, making interactive and accessible digital simulations an effective alternative. This study aims to develop an acid–base titration simulation based on Python algorithms as an innovative learning medium. The simulation focuses on the reaction between a strong acid (HCl) and a strong base (NaOH) with predetermined volume and concentration parameters, displaying pH changes through a titration curve. The research method applies an algorithmic approach based on stoichiometric calculations, visualized using the Matplotlib library. The simulation results show that the model accurately represents pH changes from acidic conditions to the equivalent point and continues toward basic conditions, consistent with analytical chemistry theory. Validation was conducted through theoretical comparison, visualization against literature curves, and expert evaluation, resulting in an average score of 3.79 out of 4. This indicates that simulation is highly feasible for use in learning. The Python-based simulation not only enhances understanding of titration concepts but also serves as an alternative learning solution in schools with limited laboratory resources.

Penerapan Algoritma K-Means Clustering Untuk Pengelompokan Data Absensi Siswa Menggunakan Teknologi QR Code Berbasis Web

2026-03-24T14:51:53+07:00

This study develops a web-based student attendance system using QR Code technology integrated with the K-Means Clustering algorithm. The system aims to improve accuracy and efficiency in managing attendance while classifying student discipline levels based on indicators of present, late, sick, excused, and unexcused. System development applies the Waterfall model. The K-Means algorithm groups students into three clusters: high, medium, and low attendance. Precision testing shows 95.56% accuracy compared to manual teacher assessment. The system supports school digitalization by providing automated attendance recording and data-driven discipline evaluation.

Implementasi Algoritma AES-256 dalam Enkripsi Pengamanan Data pada PT Kallista Prima

2026-03-17T06:05:42+07:00

This study aims to implement the Advanced Encryption Standard (AES-256) algorithm to secure pharmaceutical product pricing data within a product management system, using the AB-VAS K – TABLET product as a case study. The background of this research is based on the importance of protecting pricing data—namely HNA, HNA+PPN, and HET—from risks of manipulation and unauthorized access in information systems. The method used includes system design, encryption and decryption processes using AES-256 with a 256-bit key length and 14 transformation rounds, as well as testing data integrity and sensitivity through the avalanche effect test. The results show that all pricing data were successfully transformed into ciphertext with no recognizable pattern and could be restored identically through the decryption process. The testing also demonstrated that a small change in plaintext produced a significant change in the resulting ciphertext, thereby increasing the security level of the system. With a key complexity of 2²⁵⁶, the implementation of AES-256 is considered effective in maintaining the confidentiality and integrity of pricing data and is suitable for application in pharmaceutical product management information systems.

Perancangan Sistem Informasi Manajemen Kelelahan Driver Berbasis Web pada Honda Amartha Samarinda dengan Metode Self-Assessment dan Penjadwalan Delivery

2026-03-20T07:38:26+07:00

Kelelahan kerja pada driver merupakan faktor yang memengaruhi keselamatan berkendara, kualitas pelayanan, dan kinerja operasional Perusahaan. Pada Honda Amartha Samarinda, pemantauan kelelahan driver dan penjadwalan delivery masih dilakukan secara manual sehingga berpotensi menimbulkan ketidakefisienan dan kesalahan pencatatan. Penelitian ini bertujuan merancang Sistem Informasi Manajemen Kelelahan Driver berbasis web dengan metode self-assessment dan penjadwalan delivery untuk meningkatkan efektivitas monitoring dan distribusi tugas. Metode penelitian meliputi analisis kebutuhan sistem, perancangan menggunakan Unified Modeling Language (UML), perancangan basis data, dan implementasi sistem berbasis web. Metode self- assessment diterapkan melalui pengisian kuesioner oleh driver sebelum melakukan delivery untuk mengidentifikasi tingkat kelelahan, yang kemudian menjadi parameter dalam penjadwalan tugas secara terintegrasi. Sistem yang dirancang mampu menyajikan informasi tingkat kelelahan secara real-time, sehingga dapat membantu admin logistik dalam menentukan kelayakan penugasan, serta menghasilkan laporan monitoring sebagain bahan evaluasi manajemen. Hasil penelitian menunjukan bahwa sistem ini dapat meningkatkan efisiensi pengelola jadwal, mendukung pengambilan keputusan, serta berpotensi meningkatkan keselamatan dan produktivitas kerja driver.

Perancangan dan Implementasi Aplikasi E-Library Berbasis Web yang Berpusat pada Pengguna

2026-04-09T12:03:29+07:00

Advances in information technology have driven the transformation of library services from conventional systems to more efficient and accessible digital systems. However, digital libraries still face various challenges, such as limited access, a lack of personalized services, and low interoperability and accessibility for users with special needs. This study aims to design and develop a web-based e-library application capable of improving the effectiveness of library management and providing a more adaptive and inclusive information access experience. The methodology used is Design Science Research (DSR), which includes the stages of problem identification, needs analysis through literature review, interviews, and questionnaires, system design using a human-centered design approach, and system prototype development. System design was conducted using Unified Modeling Language (UML) modeling, consisting of use case diagrams, activity diagrams, and class diagrams. The developed system provides user authentication, book search, access to digital resources, and collection and user management by administrators. System testing was conducted using the Black Box Testing method to ensure that every function operates according to specifications. The research results indicate that the developed e-library application is capable of functioning effectively and supports the management of digital libraries more efficiently.

Penerapan Sistem Persediaan Berbasis Microsoft Excel sebagai Solusi Peningkatan Efisiensi Pengelolaan Bahan Baku pada Perusahaan Konveksi di Lumajang

2026-04-02T22:10:23+07:00

Penelitian ini bertujuan untuk menganalisis penerapan sistem persediaan berbasis Microsoft Excel dalam meningkatkan efisiensi dan Ketepatan Data pengelolaan bahan baku pada Konveksi di Lumajang. Perusahaan menerapkan sistem produksi berdasarkan pesanan dengan spesifikasi yang beragam, namun belum memiliki pencatatan penggunaan bahan baku secara rinci per pesanan sehingga menyulitkan dalam mengetahui pemakaian bahan, sisa stok, dan perhitungan harga pokok produksi secara akurat. Metode penelitian yang digunakan adalah pendekatan kualitatif deskriptif dengan teknik pengumpulan data berupa observasi, wawancara, dan dokumentasi. Hasil penelitian menunjukkan bahwa penerapan sistem berbasis Microsoft Excel yang terdiri dari master data, lembar transaksi, dan rekap stok otomatis mampu meningkatkan akurasi, efisiensi, serta kemudahan dalam pemantauan persediaan. Sistem ini berkontribusi dalam meningkatkan kemampuan Perusahaan dalam mengendalikan penggunaan bahan baku dan mendukung pengambilan keputusan secara lebih tepat. Namun, efektivitas sistem bergantung pada kedisiplinan pengguna dalam melakukan input data serta keterbatasan Excel dalam integrasi sistem.

Evaluasi Model Convolutional Neural Network (CNN) dalam Klasifikasi Penyakit Daun Jagung Berbasis Web Menggunakan Citra Digital

2026-03-22T08:07:41+07:00

Corn leaf diseases such as blight and rust can reduce crop yields if they are not detected at an early stage. In Penatahan Village, the process of identifying these diseases is still carried out manually through visual observation, which may lead to misidentification due to the similarity of symptoms between different diseases. Therefore, a technology-based system is needed to assist the identification process in a more objective and efficient manner. This study aims to classify corn leaf diseases using the Convolutional Neural Network (CNN) method based on digital leaf images. The dataset used consists of 319 images categorized into three classes: healthy, blight, and rust, with 80% of the data used for training and 20% for validation. The model was developed using a transfer learning approach with the MobileNetV2 architecture and evaluated using a confusion matrix. The experimental results indicate that the model achieved an accuracy of 92.19%, indicating that the CNN method is capable of effectively classifying corn leaf diseases. The developed system can be utilized as a tool to assist in the rapid and objective identification of corn leaf diseases.

Hybrid Feature Selection dan Ensemble Learning untuk Klasifikasi Risiko Stunting Anak di Indonesia

2026-04-15T12:43:36+07:00

Stunting is a chronic nutritional problem that remains a major public health issue in Indonesia. This study aims to develop a classification model for stunting risk in children using a combination of hybrid feature selection and ensemble learning methods. The dataset used is derived from socio-economic and health data obtained from the Central Statistics Agency and open datasets. The research method includes data preprocessing, feature selection, model development using Random Forest and Gradient Boosting combined with a Voting Classifier, and evaluation using accuracy, precision, recall, F1-score, and ROC-AUC metrics. The results show that the proposed model achieves high performance with accuracy reaching 98% and ROC-AUC close to 1. The hybrid feature selection successfully improves model efficiency by selecting relevant features. This study demonstrates that the integration of feature selection and ensemble learning can produce an accurate and interpretable model for early detection of stunting risk.

Perancangan Knowledge Management System Berbasis Website Menggunakan Model SECI untuk Mendukung Knowledge Sharing Guru pada SMP Bina Karya

2026-04-13T14:12:14+07:00

The design of a website-based Knowledge management system (KMS) using the SECI model at SMP Bina Karya is motivated by several problems, including knowledge that remains stored individually with each teacher, the unavailability of centralized learning materials and lesson plans (RPP), and the difficulty faced by substitute teachers in delivering lessons when replacing the main teacher who is absent. The Knowledge management system serves as a solution to document, distribute, and prevent the loss of knowledge, while also acting as a medium to enhance the culture of knowledge sharing among teachers. The design method used is a qualitative approach consisting of data collection through observation, interviews, and literature studies, identification of knowledge management using the SECI model, system requirements analysis, system design, and testing using Focus Group Discussion (FGD). This study produces a website-based KMS equipped with features such as user account management, substitute teacher schedule management, learning material management, lesson plan management, and discussion forums. The results of the FGD testing show an average acceptance rate of 94.2% for all developed features, with the substitute teacher schedule management feature serving as the main differentiator that successfully addresses the specific problems at SMP Bina Karya.

Pengembangan Sistem Informasi Deteksi Stunting Berbasis Android Menggunakan Metode Z-Score

2026-04-17T07:35:53+07:00

Stunting is one of the major chronic nutritional problems in Indonesia that affects children's growth and development. The manual process of detecting stunting often leads to delays in handling and potential inaccuracies in determining nutritional status. Therefore, a system is needed to support faster and more accurate detection. This study aims to develop an Android-based stunting detection information system using the Z-score method. The research method used is Research and Development (R&D) with the Waterfall model, which includes requirement analysis, system design, implementation, and testing stages. The data used in the calculation refers to the growth standards established by the World Health Organization (WHO). The results show that the developed system is capable of automatically calculating Z-score values based on user input, including age, height, and weight of the child. The system can also accurately determine the nutritional status according to WHO classification standards. Based on Black Box Testing, all system functions run properly as expected. Thus, the developed information system can assist the community and health workers in conducting early detection of stunting more effectively and efficiently.

Implementasi Knowledge Management Berbasis Model SECI di Perpustakaan Daerah Provinsi Sumsel

2026-04-16T14:45:56+07:00

This study analyzes the implementation of knowledge management based on the SECI model at the Regional Library of South Sumatra Province. The method used is qualitative with a case study design through a systematic literature review and digital content analysis of the website, OPAC INLISLITE, social media, and the DiarySumsel application. The results show that the implementation of the SECI model has been carried out across its four stages. Socialization is realized through direct services and mobile libraries using 4 mobile units. Externalization is evidenced by the documentation of circulation service Standard Operating Procedures (SOPs). The combination is implemented through the integration of INLISLITE and DiarySumsel, which served 23,946 users. Internalization is reflected in the adoption of digital systems by librarians and users. Supporting factors include technology availability, management commitment, and extensive service coverage. The challenges faced are limited trained human resources, suboptimal technology utilization, and low community information literacy.

Penerapan Metode Holt-Winters untuk Memprediksi Produksi Biji Kopi Arabica Lintong Nihuta

2026-04-14T21:42:19+07:00

Arabica coffee bean production in Lintong Nihuta fluctuates every month, requiring a method to predict future production volumes. Accurate predictions can aid production planning and decision-making. This study aims to predict Arabica coffee bean production using the Holt-Winters multiplicative method, which can capture trends and seasonal patterns in time series data. The data used are 60 monthly production data points from January 2021 to December 2025. The analysis process begins with determining the initial values of the level, trend, and seasonal components, followed by a smoothing process using parameters α = 0.4, β = 0.45, and γ = 0.35. Model evaluation was performed using the Mean Absolute Percentage Error (MAPE) using 2025 data as the evaluation data. The evaluation results show a MAPE value of 13.12%, indicating that the model has a good level of accuracy. The prediction results show that Arabica coffee bean production in 2026 is expected to fluctuate, with the highest predicted value in December at 75,297.15 kg and the lowest in May at 36,737.38 kg. Therefore, the Holt-Winters multiplicative method can be used to predict Arabica coffee bean production in the Lintong Nihuta District in the future.

Pengembangan Intelligent Web-Based System untuk Diagnosa Kerusakan Sepeda Motor Listrik Menggunakan Naive Bayes

2026-04-13T17:36:52+07:00

The rapid growth in the use of electric motorcycles has increased the need for a damage diagnosis system that is fast, accurate, and easily accessible. The process of identifying faults, which still relies on manual inspection, often requires considerable time and depends heavily on the expertise of technicians. This study aims to develop a web-based intelligent system capable of automatically diagnosing electric motorcycle faults using the Naïve Bayes method. This method is chosen due to its ability to perform probabilistic classification with good accuracy, even when dealing with limited data. The developed system utilizes symptom data and types of faults as the basis for probability calculations to determine the most likely damage. The system development process includes data collection, model design, algorithm implementation, and system testing. The testing results indicate that the system is able to provide fast and consistent diagnostic recommendations based on the symptoms input by users. With the implementation of this system, it is expected to assist both users and technicians in conducting initial identification of electric motorcycle faults more efficiently, thereby accelerating the repair process and reducing the potential for diagnostic errors.

Perancangan Aplikasi Mobile untuk Rekomendasi Kuliner Lokal

2026-04-21T11:54:45+07:00

The rapid advancement of information technology has shifted the paradigm of tourism and culinary exploration towards mobile platforms. However, local culinary searches in Medan City are currently still conventional and random. Scattered culinary information across various social media without a comprehensive summary often confuses customers and complicates the promotion of micro-culinary businesses (UMKM) in the digital era. Existing commercial platforms often burden local sellers with high platform taxes. This research aims to design and implement an Android-Based Mobile Application for Local Culinary Recommendations as an effective solution to these problems. This application is built natively using the Java programming language for the Android client side, integrated with a PHP Application Programming Interface (API) and MySQL relational database. The system design involves a Client-Server architecture and includes three distinct access rights: User, Culinary Admin, and Super Admin. Implemented features include dining place search, interactive reviews, ratings, popularity levels based on view counts, and Google Maps API integration for precise culinary navigation routes. The result of this research is a self-managed mobile application that accelerates navigation access for users while assisting business owners in managing their stalls independently, objectively, and expanding their promotional reach without internal transaction fees.

KUTA-TRANSPARAN: Model Sistem Informasi Terintegrasi untuk Pencegahan Pungutan Liar pada Desa Wisata Berbasis Fraud Risk Management dan Akuntansi Digital

2026-04-16T16:24:53+07:00

Illegal levies (pungli) in tourism village public services threaten transparency and accountability of village governance. This study develops the KUTA-TRANSPARAN Framework as an integrated information system model for preventing pungli, with a case study in Kutagugung Village, Karo Regency, North Sumatra. The research adopts Participatory Action Research (PAR) integrated with Design Science Research (DSR), the GAO Fraud Risk Management Framework, and the Technology Acceptance Model (TAM), with hypothesis testing using Structural Equation Modeling–Partial Least Squares (SEM-PLS) on a sample of 120 respondents (purposive sampling). The framework encompasses three dimensions: (1) technology, based on mobile applications and blockchain; (2) business processes, covering prevention (Electronic Village Services/EVS, Electronic Tourism Retribution/ETR, Digital BUMDes Accounting/DBA), detection (real-time monitoring, whistleblowing), and response (auto-sanctions); and (3) institutional, covering policy and transparency culture. The ETR module achieved 81.3% transaction time efficiency (reduced from 15–30 minutes to 2.8 minutes), while the EVS module achieved 70% administrative processing efficiency (reduced from 3–7 days to 4.2 hours). Pilot results show a 23% increase in village retribution revenue (from IDR 12.3 million to IDR 15.1 million per month) attributable to the elimination of revenue leakage. All five hypotheses were supported, with the integration of management information systems and digital accounting demonstrating the strongest effect on financial accountability (β=0.79; p<0.001). The theoretical contribution lies in an interdisciplinary framework integrating informatics management and accounting computerization for fraud prevention in tourism villages. Practically, the blueprint supports SDGs Goal 16 and Goal 17.

Implementation of the SMART Method in a Decision Support System for Scholarship Selection Using Real Data

2026-04-11T12:53:38+07:00

The process of selecting scholarship recipients at SMK MEDAN still faces issues of subjectivity and inefficiency due to manual procedures. This study aims to develop a scholarship decision support system based on the SMART (Simple Multi Attribute Rating Technique) method within the Decision Support System (DSS) framework to enhance objectivity and transparency in decision-making. This research employs a quantitative approach within an applied research framework using real student data based on criteria such as academic achievement, parents’ income, number of dependents, and attendance. The results indicate that the SMART method effectively generates a consistent multi-criteria ranking model, where the top candidate achieves a preference score of 0.833, equivalent to an 83.3% eligibility level. The developed system improves decision quality through more objective and efficient educational decision optimization. The main contribution lies in integrating SMART with real data into an application-based system, offering a more transparent and practical approach compared to conventional methods.

Perbandingan Algoritma Naive Bayes dan K-Nearest Neighbors dalam Analisis Sentimen Ulasan Aplikasi Gojek

2026-04-26T16:47:15+07:00

Sentiment analysis of mobile app reviews helps understand public perceptions of digital service quality. This study compares two machine learning algorithms, Naive Bayes (NB) and K-Nearest Neighbors (KNN), for classifying sentiment in Gojek app reviews from the Google Play Store. The dataset includes 5,000 reviews (1-star as negative and 5-star as positive), processed through Indonesian text preprocessing steps: case folding, tokenization, stopword removal, stemming with PySastrawi, and TF-IDF feature extraction using unigrams and bigrams. After cleaning, 4,685 valid reviews remained, split into 80% training and 20% testing, producing 3,322 features. Results show that Naive Bayes (MultinomialNB, α = 1.0) outperforms KNN, achieving 89.43% accuracy, 90.09% precision, 89.43% recall, and 89.34% F1-score, with a 5-fold cross-validation score of 91.22%. Meanwhile, KNN (k = 7, cosine metric) achieves 86.77% accuracy, 86.78% precision, 86.77% recall, and 86.75% F1-score, with a cross-validation score of 87.83%. Overall, Naive Bayes proves more effective for high-dimensional Indonesian text classification using TF-IDF.

Penerapan Association Rule Menggunakan Algoritma Apriori untuk Rekomendasi Strategi Penjualan pada UMKM Toko Pempek

2026-04-24T16:04:29+07:00

Pempek Ceria SME has a growing number of daily sales transactions; however, these data have not been optimally utilized to support sales strategies. This situation highlights the need for transaction data analysis to understand customer purchasing patterns and develop more effective promotional strategies. Therefore, this study focuses on applying association rules using the Apriori algorithm to provide sales strategy recommendations for Pempek Ceria SME. The analysis was conducted using RapidMiner software on 317 transactions from October to December 2025, with a minimum support of 15% and a minimum confidence of 65%. The results show two association rules that meet these criteria: the combination of Pempek Adaan and Orange Juice, with a support of 28% and confidence of 72%, and the combination of Pempek Kapal Selam and Sweet Iced Tea, with a support of 27% and confidence of 70%. These findings indicate that the association rule method based on the Apriori algorithm can identify relationships between menu items frequently purchased together. By understanding these purchasing patterns, Pempek Ceria SME can optimize bundling strategies and product recommendations to improve promotional effectiveness and sales.