Journal of Applied Research In Computer Science and Information Systems

Comparison of Apriori and Fp-Growth Algorithms in Determining Package Menus at Sate Perawan Restaurant Sawangan Raya

Shabrina Putri, Ninuk Wiliani, Febri Maspiyanti — Mon, 30 Dec 2024 00:00:00 +0000

The culinary creative industry holds promising prospects as it is a necessity for society. However, the variety of menu items and high customer demand lead to slow ordering processes, which hinder service at Rumah Makan Sate Perawan. Additionally, some menu items are less popular among customers. To address these issues, a system is needed to assist in determining food and beverage package menus based on association rules. This system aims to facilitate business owners in organizing packages and improving sales. This study employs the Apriori and FP-Growth algorithms, using sales transaction data collected over a four-month period. The research applies a minimum support of 0.1 for food, 0.01 for beverages, and a minimum confidence of 0.6 for both categories. The results indicate that there is no significant difference between the two algorithms in terms of the generated packages, lift ratio evaluation, and runtime. In the food category, 5 association rules were generated with an average lift ratio of 1.1929, while in the beverage category, 2 rules were generated with an average lift ratio of 1.8990.

Enhancing Text Classification Performance: A Comparative Study of RNN and GRU Architectures with Attention Mechanisms

Mon, 30 Dec 2024 00:00:00 +0000

Text classification plays a crucial role in natural language processing, and enhancing its performance is an ongoing area of research. This study investigates the impact of integrating attention mechanisms into a recurrent neural network (RNN) based architectures, including RNN, LSTM, GRU, and their bidirectional variants (BiLSTM and BiGRU), for text sentiment analysis. Three attention mechanisms Multihead Attention, Self Attention, and Adaptive Attention are applied to evaluate their effectiveness in improving model accuracy. The results reveal that attention mechanisms significantly enhance performance by enabling models to focus on the most relevant parts of the input text. Among the tested configurations, the LSTM model with Multihead Attention achieved the highest accuracy of 68.34%. The findings underscore the critical role of attention mechanisms in overcoming traditional RNN limitations, such as difficulty in capturing long-term dependencies, and highlight the potential for their application in broader text classification tasks.

Customer Segmentation of Cash Management System Using K-Means Clustering

Rizki Hesananda, Patri Apriliga — Mon, 30 Dec 2024 00:00:00 +0000

The effective financial management is essential for running successful business operations. In the banking context, the Cash Management System (CMS) facilitates real-time, automated transaction management. PT Bank Rakyat Indonesia (Persero) Tbk., as one of Indonesia’s largest banks, has implemented CMS since 2009. Despite its benefits, challenges persist, such as customer transactions outside regular working hours and difficulties in segmenting customers based on transaction volume and frequency. This study aims to address these issues by clustering BRI CMS users using the K-Means Clustering method, following the CRISP-DM framework. The research utilized transaction data of 2,727 users from January 2021 to April 2022. Data preparation involved cleaning anomalies and converting non-numeric values to numeric formats. Using the Elbow method, the optimal number of clusters was determined, resulting in three distinct user segments. The clustering revealed actionable insights, such as identifying high-value customers for targeted marketing and improving service strategies. This research offers a novel application of K-Means Clustering and CRISP-DM to CMS data management, contributing to better customer segmentation and strategic decision-making. These findings can help banks optimize resources, improve customer satisfaction, and enhance overall transaction efficiency.

Fruit and Vegetable Classification using Convolutional Neural Network with MobileNetV2

Muhammad Khoiruddin, Silvester Tena — Mon, 30 Dec 2024 00:00:00 +0000

Fruits are parts of plants that originate from the plant's pistils and usually contain seeds. Meanwhile, vegetables are leaves, legumes, or seeds that can be cooked. Fruits and vegetables have many variations that can be distinguished based on color, shape, and texture. However, the development of Artificial Intelligence (AI) technology has become pervasive in everyday life, one aspect of which is demonstrated through deep learning, a method of AI learning. Therefore, developing deep learning for tasks such as automatically detecting surrounding objects is necessary. This study aims to classify types of fruits and vegetables by applying a Convolutional Neural Network (CNN) with the MobileNetV2 architecture. In this study, fruits and vegetables encompassing 36 categories, including significant types in daily life, were considered. The results show that the classification system achieved an excellent accuracy rate of 97.31%, demonstrating the effectiveness of using deep learning techniques for this application

Identifying Damage Types in Solar Panels Through Surface Image Analysis with Naive Bayes

Ninuk Wiliani, Titik Khawa Abdul Rahman, Suzaimah Ramli — Mon, 30 Dec 2024 00:00:00 +0000

The growing utilization of solar panels as a renewable energy source requires efficient maintenance solutions to guarantee their best functioning. Identifying and categorizing faults on solar panel surfaces is essential for maintenance, as these defects considerably affect energy output and system efficiency. This study investigates the utilization of statistical feature extraction methods alongside Bernoulli Naive Bayes (BNB) and Gaussian Naive Bayes (GNB) algorithms to categorize different defect types, such as cracks, scratches, spots, and non-defective surfaces, through digital image analysis. Statistical criteria, including recall, specificity, and area under the curve (AUC), are employed to assess model performance. The findings indicate that the GNB algorithm surpasses BNB, with a mean average precision (mAP) of 39.83% with an 85:15 training-test ratio, whereas BNB reaches a maximum mAP of 29.25% at a 90:10 ratio. Nonetheless, both models demonstrate constraints in precision, as indicated by a total AUC of 0.644. This work illustrates the potential of statistical feature extraction approaches for defect classification, while emphasizing the necessity for future improvements to boost the efficacy of feature extraction and classification techniques in practical applications

Integrated Renewable Energy Microgrid Model Based on Blockchain and DeFi: A Simulation Study of P2P Energy Trading and Renewable Energy Certificate Tokenization

erick fernando, Marojahan , Fahmy Rinanda Saputri — Mon, 30 Dec 2024 00:00:00 +0000

This study proposes and evaluates a blockchain-based renewable energy microgrid model and Decentralized Finance (DeFi) that integrates physical energy systems, IoT metering, smart contracts, peer-to-peer (P2P) energy markets, Renewable Energy Certificate (REC) tokenization, and staking mechanisms to create a transparent and incentivized energy-finance ecosystem. The evaluation is conducted through hourly computational simulations for one year (8,760 hours) on the same community microgrid configuration, namely 500 kWp PV and 200 kWh batteries, with a total consumption of 1.25 GWh. Three scenarios are compared to isolate the impact of the digital layer on system performance: S1 (baseline without blockchain and without P2P), S2 (blockchain-P2P with energy trading and RECs without DeFi), and S3 (full integration of blockchain + DeFi). The results show that S1 results in 36% renewable energy penetration with 810 MWh/year of grid energy imports and a system cost of IDR 455 million/year, indicating that the utilization of PV surplus and the role of batteries is still limited. In S2, the implementation of a P2P marketplace and REC tokenization increased renewable energy penetration to 52%, decreased grid imports to 600 MWh/year, and reduced system costs to IDR 382 million/year, due to increased battery utilization and reduced curtailment. In S3, the DeFi staking mechanism (10%/year yield) strengthened green energy utilization incentives, increasing renewable energy penetration to 67%, decreasing grid imports to 430 MWh/year, and decreasing net system costs to IDR 305 million/year after incorporating DeFi revenue of approximately IDR 48 million/year, with stable tokenomics indicated by approximately 55% of tokens being staked. These findings confirm that the gradual integration of blockchain and DeFi can improve the technical and economic efficiency of microgrids, while transforming renewable energy into a productive digital asset.