Journal of Applied Intelligent System

Artificial Intelligence Chatbot for Customer Service in E-Commerce Using Telegram Based on Node.js

Fuaida Nabyla — 2024-08-14

Currently, the traditional market is increasingly being supplanted by numerous online markets. The fierce competition in the online market necessitates excellent service from sellers to buyers. As a result, many online stores now offer round-the-clock service, which can be financially burdensome if handled manually. Chatbots offer a promising solution by automating the online shopping process, thereby reducing costs and enhancing customer service. To address the need for accurate and prompt responses, this study proposes an intelligent chatbot system built on Artificial Intelligence (AI), specifically tailored to function as an e-commerce assistant. Integrated seamlessly into the Telegram application, the chatbot efficiently processes user input questions through three essential stages: parsing, pattern matching, and data crawling, all powered by AI technology. Within the AI process, user requests are systematically categorized into three primary domains: general questions, calculations, and stock checks. Notably, the calculation category encapsulates both order and payment processes. The effectiveness of the proposed method is substantiated by the results of 200 trials, demonstrating its adeptness in accurately addressing all user inquiries.

Smart Waste Management and Recycling Based on IoT using Machine Learning Algorithm

Gerinata Ginting — 2024-08-14

Smart waste management and recycling have become critical issues in urban planning and environmental sustainability due to the increasing volume of waste generated by modern societies. In this study, we investigated the performance of Support Vector Machine (SVM) and Neural Network (NN) methods in an Arduino-based waste sorting system. Our testing phase revealed exceptional performance, with SVM achieving an accuracy of 92% and NN achieving 96%, alongside perfect precision, recall, and F1-score metrics. The consistent True Positive (TP) results across all waste categories underscored the system's capability to accurately direct waste into correspondingcolored bins. These findings highlight the significance of automated waste management systems in promoting effective waste sorting practices and facilitating sustainable recycling efforts. Moreover, advancements in technology and machine learning algorithms offer promising prospects for further enhancing the efficiency and scalability of such systems, thereby contributing to a cleaner and healthier environment for future generations. Future research in smart waste management could focus on exploring additional machine learning algorithms, advanced sensor technologies, and Internet of Things integration. Investigating alternative algorithms beyond SVM and NN, adopting advanced sensors like hyperspectral imaging or lidar, and incorporating IoT devices for real-time monitoring could enhance waste sorting accuracy and scalability.

Pneumonia Detection on X-rays Image using YOLOv8 Model

Agoes Santika Hyperastuty — 2024-08-14

Pneumonia is an acute inflammatory disease of lung tissue. It is usually caused by microorganisms such as bacteria, fungi and viruses. The young children are particularly vulnerable to this illness. Report in 2019 shows that pneumonia kills almost 2,000 children under the age of five every day worldwide and affects over 800,000 children under the age of five annually. Analyzing the chest X-ray results of the patient's body is one method of diagnosing pneumonia. Therefore, this research was done to deploy a deep learning to identify the healthy and pneumonia affected lungs from chest X-ray images in order to aid in the diagnosing process. This research was done by using 2000- chest X-ray dataset—of which 1500 pneumonia lung data and 500 normal lung data. The computer vision model YOLOv8 is used in this study. The accuracy results from the training process were 56.15% in the pneumonia class and 92.03% in the normal class. Wether in the testing process yielded an average value of 0.482 (48, 2%) for the pneumonia class and 0.675 (67,5%) for the normal class. From these results, there are promising possibilities for developing a pneumonia detection system using YOLO in the future.

K-MEANS ALGORITHM IN CLUSTERING SALES DATA FOR CALCULATING ESTIMATED HOUSE PRICES

Gatot Tri Pranoto — 2024-08-14

Determination of the value of the guarantee to the Bank in the process of applying for Home Ownership Credit (KPR) submitted by prospective customers still refers to the provisions of the Financial Services Authority, where the assessment must follow the existing rules and be carried out by public appraisals or commonly called the Office of Public Appraisal Services (KJPP). Currently the analyst credit officer cannot validate the results of the assessment report from KJPP, so if an error occurs either intentionally or not by KJPP or appraisal parties continue to process according to the given value. In the event of default of payment by the customer due to the lower collateral value of the loan provided, the bank violates Bank Indonesia Regulation number 18/16/PBI/2016 concerning loan to value ratio. This study aims to apply the K-Means algorithm in grouping home sales so that it can be used for the calculation of the estimated value of house prices, and develop a prototype of the house price estimation information system. Data retrieval using crawling or scrapping techniques on the website makes it easier to fulfill data on a dataset. The result of this study is the data of home sales for kebon Jeruk area spread across the internet can be grouped into 3 clusters with the value of David Bouldin Index in duri Kepa sub area, which is 0.096, in South Kedoya sub area of 0.087, in North Kedoya sub area of 0.071, and Kelapa Dua sub area of 0.117. By combining clusterization results using K-Means methodology with land price calculation formula obtained land price estimation in sub area.

Keywords: K-Means, KPR, Data Scraping, KJPP, MAPPI

Mapping System Design of a Genetic Mapping System for the White Nest Swallow (Collocalia Fuciphaga) in Java

Patmi Kasih — 2024-08-14

The distribution of the white swift (Collocalia fuciphaga) is on several large islands in Indonesia. The white swallow has an advantage over the nests of the black swallow and sriti, having edible parts reaching 85-100% of the total nest. Many types of white swallow's nests in Indonesia are known for their different physical characteristics, color and weight. It is not known that these differences occur due to genetic differences or simply due to differences in the type of food and living environment. Differences due to food and environmental factors do not really affect the health function of the nest. Differences due to genetics greatly influence the function of the nest for health. The design of this system is initial research to start mapping white nest swiftlets in Indonesia. The results of the design will be used as a data storage system for genetic mapping of white nest swallows on the island of Java by taking DNA samples of swallows from various habitat areas. The system will store data on habitat areas, record location points and take bird samples from these areas to then carry out laboratory tests to determine the DNA code test of each bird sample. Furthermore, it is hoped that clear genetic mapping results can be used to determine the quality and function of the bioactive components of white swift nests on the island of Java. The mapping results will also be a source of knowledge about the richness of the germplasm of native Indonesian swiftlets.

Keywords: GIS, Bioactive, genetic, Java, mapping, white swallow's nest

C4.5 Algorithm Based on Forward Selection and Particle Swarm Optimization for Improving Accuracy in Heart Disease Patient Classification

Aji Awang Setiawan — 2024-08-14

Early detection of heart disease is crucial given the high number of cases occurring
in advanced stages and affecting individuals in their productive years. Utilizing data mining, the C4.5 Algorithm is one method capable of detecting the onset of heart disease, prompting timely awareness and early prevention. The dataset employed is the Heart Disease Cleveland UCI from Kaggle, featuring 13 input attributes and 1 target attribute. Using the Decision Tree method results in decision-making by constructing a decision tree. The test outcomes revealed an accuracy rate of 77.11% with the C4.5 algorithm, 83.69% with the C4.5 algorithm employing Forward Selection, and 84.73% with the C4.5 algorithm based on Forward Selection and Particle Swarm Optimization.

Vehicle Detection Using Image Conversion Percentage to Binary Method Based on K-Means

Candra Irawan — 2024-08-14

Vehicle detection is the artificial intelligence that can help us in transportation highway systems like counter vehicles passing through the road on Eid Mubarak day etc. The object in this case is divided into six classifications there are car, motorbike, van, truck, and three-wheel. On the dataset vehicle is mostly an image of a car that we get from Kaggle. To solve vehicle detection problems such as poor vehicle detection and reduced detection accuracy, we provide a new vehicle detection with a dataset at kaggle. The clustering process consists of steps in which input images are transformed into morphometrics. The next step is to classify the image data using the K-Means algorithm. The images grouped by this detection are vehicles. The first step is to determine the randomly drawn mean or center point of two image data values in the database. If there is no data transfer, the group is considered stable and group creation is completed. Seven vehicle image data are used to test this application. And the result of our experiment on vehicle detection is about 85.7 % accurate

Improving Heart Disease Severity Prediction Using SMOTE for Imbalanced Data

Ayu Hendrati Rahayu — 2024-08-14

The heart disease is a prevalent and potentially fatal condition affecting individuals worldwide. In this study, we address the challenge of predicting the severity of heart disease using supervised learning techniques. Leveraging a dataset comprising various demographic and clinical attributes, we propose a solution that employs machine learning models to accurately predict the severity level of heart disease. Among the evaluated models, Random Forest emerges as the top performer, showcasing exceptional precision, recall, accuracy, and F1-score across all severity levels, with an overall accuracy of 98.8%. This highlights the robustness of the Random Forest model in accurately classifying instances across different severity levels. Following closely behind, the KNN algorithm demonstrates commendable performance, achieving an accuracy of 92% and showcasing competitive precision, recall, and F1-score values, particularly for higher severity levels. Despite its notable aspects, XGBoost ranks third among the evaluated models, with an accuracy of 90.4%. While XGBoost excels in certain aspects, such as recall for Level 3 severity, it falls short in overall performance compared to Random Forest and KNN. For future research, exploring ensemble methods that combine the strengths of different algorithms could yield even better classification results, providing avenues for further improvement in predicting the severity of heart disease

Analysis of K-Nearest Neighbor (KNN), Naive Bayes ands Decision Tree C4.5 Algorithm With Classification Method In Breast Cancer Using RapidMiner

Muhammad Iqbal — 2024-08-14

Breast cancer is cancer that forms in the cells of the breast. It is the most common cancer in women and the leading cause of cancer deaths in women worldwide. Breast cancer is usually divided into two types: benign, or usually called benign and malignant, or usually called malignant. Benign cancers are usually characterized by small, round, tender lumps. In the fields of medicine, finance, marketing, and social science, data mining is a popular tool for performing proven analysis. This study will compare K-Nearest Neighbor (KNN), Naive Bayes, and Decision Tree C4.5 approaches for classifying breast cancer. The problem of this research is which algorithm has a high level of accuracy that can be used with breast cancer datasets and can provide information about patterns or models for early detection of breast cancer. The results of the research conducted using CRISP-DM show that K-Nearest Neighbor (KNN) has the highest accuracy value with 97.14% and its AUC value is 0.976. The AUC value also showed excellent classification, with an AUC value between 0.90 and 1.00.

English Class Scheduling Information System at Indonesian-American Educational Institutions

Faik Bajsair — 2024-08-14

The purpose of the research is to create and implement a simple class scheduling application that is useful to minimize the occurrence of clashes of time, classes, levels, teachers and students at the same time. The research method used is the Descriptive Method with the type of case study research. The descriptive method is a method of researching the status of a group of people, an object, a set of conditions, a system of thought or an event in the present. From this Thesis or Final Project, the author can draw the conclusion that the English Class Scheduling Information System in Indonesian-American Educational Institutions is more effective, fast, conceptual, and up to date in data processing

Comparison of Shallot Price Prediction In Pati City With LSTM, GRU and Linear Regression

Fajar Husain Asyari — 2024-08-14

Shallots are superior vegetable plant and contribute quite significantly to the development of the national economy. The price of shallots fluctuates almost every year. At certain times the price of shallots soars due to high demand while the supply in the market is insufficient. Therefore, an analysis is needed to see what phenomena significantly affect the increase in the price of shallots. The methods used in the study were LSTM, GRU and LR. The results of the analysis show that the LSTM algorithm gets a MAE value of 0.011072172783, MAPE 3.93678% and RMSE 0.03139695060, this error is the lowest compared to GRU getting MAE value is 0.01185741, MAPE 4.2282357% and RMSE 0.03122299395 and LR with MAE 0.0134737280395416, MAPE 5.45081% and RMSE is 0.0313332635305961, so LSTM is a suitable algorithm for predicting shallot data in Pati district.