Infact: International Journal of Computers

Data-Driven Classification of Poverty Status in Indonesia using Machine Learning Techniques

2025-09-11T12:24:17+07:00

This study explores the use of the K-Nearest Neighbor (KNN) algorithm to classify poverty status in Indonesia using publicly available socio-economic indicators. Traditional poverty classification methods are often inefficient and lack nuance. By leveraging the Knowledge Discovery in Databases (KDD) process, including data preprocessing, normalization, and dimensionality reduction via PCA, the study builds a robust classification model. The dataset includes indicators such as education, health, and expenditure levels from 514 districts/cities. The optimal KNN model, determined through cross-validation, achieved a test accuracy of 95.15%, with strong precision, recall, and ROC AUC scores. Feature importance analysis via Random Forest on PCA-transformed data highlights the predictive influence of certain component combinations. The results demonstrate the potential of machine learning to support more accurate and data-driven policy targeting in poverty alleviation. Future enhancements may involve integrating time-series or satellite data to increase relevance and precision.

Prototype of an Automated Item Sorting System Using a Barcode Scanner and Servo-Based Directional Control with Microcontroller

2026-04-21T08:09:07+07:00

The increasing demand for efficiency in logistics and manufacturing highlights the limitations of manual sorting systems, which are prone to errors and inefficiency under high-volume conditions. Existing automated sorting systems often rely on multiple sensors or complex configurations, resulting in higher costs and system complexity. This study aims to develop a cost-effective and simplified automated sorting prototype using a single barcode scanner integrated with servo-based directional control. The system is designed using an Arduino Mega 2560 microcontroller, GM66 barcode scanner, infrared sensors, DC conveyor motors, and MG996R servo motors. The proposed method involves object detection, barcode identification, data processing, and directional sorting based on predefined servo angles. Experimental results show that the system successfully performs automated sorting with an overall success rate of 50%, demonstrating functional feasibility despite mechanical limitations. It can be concluded that the proposed system offers a practical and economical solution for prototype-scale automated sorting applications.

Literature Review: Utilization of Cloud Computing in Drip Irrigation System

2026-04-21T08:30:22+07:00

Low levels of food production can be caused by some countries still using traditional farming methods, which can affect crop yields. Farmers who still use conventional farming methods often cause inefficient use of resources. This can be overcome by using smart farming methods by utilize technology. This technology is known as the Internet of Things. IoT technology in agriculture is utilized in irrigation systems to create smart irrigation systems. In addition to utilizing IoT in the use of smart irrigation systems in agriculture, irrigation systems also utilize Cloud Computing. The purpose of this literature review is to examine the use of Cloud Computing in IoT-based smart irrigation systems, as well as to identify the benefits and challenges associated with efficient water use and agricultural consumption. This study uses a narrative literature review method by collecting literature studies related to the use of IoT and Cloud Computing in drip irrigation systems. From the results of the analysis of five journals discussing the use of Cloud Computing in IoT-based smart irrigation systems, it can be concluded that the use of Cloud Computing technology in smart irrigation systems provides various significant benefits, especially in terms of water use efficiency and better irrigation management.

Iot-Based Early Fire Detection System Uses MQ-2 Smoke Sensor And DS18B20 Temperature Sensor

2025-08-14T17:23:19+07:00

Fire disasters remain a major threat in Indonesia, especially in dense housing, offices, and industrial zones. Delayed detection leads to heavy property losses and fatalities, since blazes are often noticed only after flames grow and smoke spreads. This study introduces an IoT early warning system combining an MQ-2 smoke sensor and a DS18B20 temperature sensor on a NodeMCU ESP8266. Using the ADDIE model—analysis, design, development, implementation, evaluation—the prototype was built and tested in laboratory simulations. Tests show the MQ-2 detects smoke at ?400 ppm, while the DS18B20 measures temperatures ?60 °C with ±0.5 °C precision. The dual-sensor setup delivers over 95 % accuracy, alerts within two minutes, and keeps false alarms below 5 %, providing an effective and economical tool for urban fire mitigation. Its low-cost components and Wi-Fi connectivity enable real-time alerts to smartphones or control rooms, facilitating response and scalable deployment in communities.

Intelligent Service Quality Asse Aspect-Based Sentiment and Emotion Analysis on Online Reviews Using DistilBERT Method for Service Quality Evaluation

2026-04-22T11:36:16+07:00

The growth of online reviews on digital platforms has made consumer opinions an important source for understanding perceptions of service quality in businesses. This study aims to analyze aspect-based sentiment and emotion from consumer reviews using the Distilled Bidirectional Encoder Representation from Transformers (DistilBERT) method. Data were collected from Google Reviews and processed through text preprocessing, aspect extraction, sentiment and emotion labeling, and fine-tuning of the DistilBERT model. Sentiment analysis was classified into three classes (positive, negative, and neutral), while emotion analysis included five categories (happy, angry, disappointed, sad, and neutral). The evaluation results show that the DistilBERT model achieved excellent performance in sentiment classification with an accuracy of 95.00%, precision of 93.60%, recall of 95.00%, and F1-score of 94.22%. For emotion classification, the model achieved an accuracy of 94.00%, precision of 88.36%, recall of 94.00%, and F1-score of 91.09%. These findings indicate that a Transformer-based approach is effective in understanding the contextual meaning of consumer reviews despite the use of a relatively limited dataset. This study concludes that DistilBERT is capable of providing accurate and efficient aspect-based sentiment and emotion analysis, which can be utilized as a foundation for evaluating and improving service quality and digital business reputation.

Implementation of Latent Dirichlet Allocation Topic Modeling and VADER on Aspect-Based Sentiment Analysis

2026-04-22T07:55:10+07:00

Aspect-Based Sentiment Analysis on a Product or Service is Crucial for Enhancing Customer Satisfaction. This Study Applies Latent Dirichlet Allocation (LDA) Topic Modeling to Identify Aspects. Then, the Valence Aware Dictionary and Sentiment Reasoner (VADER) Lexicon Method is Adopted to Determine Sentiment on These Aspects. The Data Source Comes from Customer Reviews of a Gelato Ice Cream Shop at Taman Siswa. Data was collected from Google Maps Using the Web Scraping Method via the Instant Data Scrapper Application. The Experimental Results Show that the LDA Method Identified 3 Aspects: Flavor, Place, and Service. Meanwhile, Sentiment Measurement Using VADER on the Flavor Aspect Revealed a Positive Sentiment of 213%, Negative Sentiment of 60%, and Neutral Sentiment of 218%. The Next Aspect, Place, Had a Positive Sentiment of 32%, Negative Sentiment of 4%, and Neutral Sentiment of 4%, while the Service Aspect Composed of 32% Positive Sentiment, 3% Negative Sentiment, and 3% Neutral Sentiment. Overall, the Positive Sentiment for the Flavor Aspect (Taste) Outweighed the Negative and Neutral Sentiments for the Place (Location) and Service (Service) Aspects.