Jurnal Informatika: Jurnal Pengembangan IT
Not a member yet
437 research outputs found
Sort by
Usability Sentiment Analysis Menggunakan Metode SUMI, NLP Scikit-Learn pada Aplikasi New Sakpole
No full textThis research will discuss issues related to how to evaluate the usability and Sentiment Analysis aspects of the New Sakpole application system, how to determine the level of user satisfaction in using the New Sakpole mobile application and to determine sentiment analysis based on the results of analysis using the SUMI and NLP tools. The research objective is based on the formulation of existing problems to provide usability aspect values for the development of the New Sakpole mobile application and generate recommendations for improvement and determine the level of positive and negative sentiment analysis by using the New Sakpole Application as a medium for paying Motor Vehicle Tax. The test uses the Software Usability Measurement Inventory (SUMI) tool, the New Sakpole mobile application system, which is very helpful and can provide value to the community in the online vehicle tax payment process. This can be seen and obtained from a scale of helpfulness and efficiency resulting from a maximum score of 100 with an average score of 101 and 86.2. The results of the test using the SUMI tool, all average aspects get above average results, so the level of usability that occurs is that the use of New Sakpole has worked and is running well. The test uses Scikit-Learn Natural Language Processing (NLP) that the results of processing the review dataset on the New Sakpole Application from the Google Play Store with a total of 4704 reviews and a sampling of 500 reviews, that the response or reviews of the community using the New Sakpole application are negative even though for Acuracy word (words) that conveyed a review of 80.90%. From the results of the sample data test that index 0 is negative so that the words "good, very enlightening" can be concluded with Sentiment is 1 (POSITIVE)"
Analisis Sentimen Terhadap Calon Presiden Indonesia 2024 dengan Metode Extreme Gradient Boosting (XGBOOST)
No full textIn 2024, Indonesia will implement democracy in the election of the Indonesian head of state. Any political figure who runs for head of state and calculates his popularity based on public opinion. After the General Election Commission (KPU) released the names of the 2024 Indonesian presidential candidates, these names were widely discussed, especially on social networks, one of which was Twitter. Twitter or what is often called X is a platform that provides short, concise and clear information. Twitter users responding to the 2024 presidential candidate have different opinions on Twitter. The sentiments used are positive, negative and neutral. The method used to analyze public opinion with data processed on Twitter social media uses Extreme Gradient Boosting (XGBOOST), classifying tweet test data in the form of classification with prediction output with accurate values. This research takes Twitter data to see public opinion on presidential candidates. The aim of this research is to determine the process of digital text analysis and the application of the XGBOOST method to Twitter user sentiment in two categories (positive and negative) and three categories (positive, negative and neutral) for each candidate, namely Ganjar Pranowo, Anies Baswedan and Prabowo Subianto. The results show an accuracy of 0.96%, precision of 0.96% and recall of 0.97%
Analisis Spam Komentar Instagram menggunakan Support Vector Machine dengan Variasi Hyperparameter
No full textInstagram (IG) is a web and mobile-based social media application where users can share photos or videos with the available features. These features include captions, tagging, adding locations where photos or videos were taken, editing and filtering photos or videos before they are uploaded from the smartphone application and certain tags so that the photos can be seen by many people. Instagram as social media is not only a medium for communication but also for developing brands and selling products. Spam that often appears in spam comments is a barrier to getting appropriate information. When identifying spam and non-spam comments, a challenging problem is that the number of spam comments is less than non-spam comments, thus causing an imbalanced dataset problem. Imbalanced data sets can affect the performance of classification algorithms. Support Vector Machine (SVM) to classify comments between two classes (spam or nonspam) which is the maximum distance between the hyperplane and the closest item from both classes. Analysis of related research that has been carried out with feature variations states that the addition of 90 different features to the data used to increase classification accuracy on imbalanced data. Other related research discusses Complementary Naïve Bayes which can be used to balance dataset classes. This research describes the selection of Support Vector Machine hyperparameters, especially for unbalanced data where the level of similarity is almost the same, so hyperparameter experiments are needed for the best accurac
Performance and Security Analysis of Lightweight Hash Functions in IoT
No full textThe rapid proliferation of Internet of Things (IoT) devices across various sectors, including healthcare, automotive, smart homes, and agriculture, has created a need for robust security measures that do not compromise the limited resources of these devices. This study analyses the performance and security of several lightweight cryptographic hash functions, specifically SHAKE128, BLAKE2s, SHA-256, SHA3-256, SipHash and xxHash, within the context of the Internet of Things (IoT). A series of experiments conducted on the Arduino Uno platform allows for an evaluation of these functions in terms of throughput, memory usage, and avalanche effect. The findings indicate that while SHAKE128 and SHAKE256 demonstrate superior throughput, they require greater memory, particularly with larger input sizes. BLAKE2s exhibits a robust equilibrium between throughput, memory efficiency, and consistent avalanche effects, rendering it a dependable option for 256-bit outputs. Conversely, xxHash and SipHash provide high throughput and minimal memory usage, yet exhibit reduced avalanche effects. The findings of this research provide critical insights for developers and researchers on the selection of appropriate cryptographic solutions, which must be tailored to the constraints and security requirements of IoT devices
Aplikasi Pemandu Wisata Berbasis Android Untuk 10 Wisata Bali Baru
No full textTourism has a very significant role in the Indonesian economy. In 2016, the Indonesian government prioritized 10 tourism destinations as "10 New Bali Tourism". However, tourists often face difficulties in finding accurate and efficient information about public facilities at tourist attractions. To overcome this problem, this research aims to develop an Android-based tourist guide application. This application is designed to help tourists find facility information quickly and accurately. This research was carried out through a series of stages, including design, analysis, design, implementation, testing and application maintenance. The result of this research is the development of a tour guide application that is useful for tourists. This application makes it easy for them to find accurate information about public facilities around the tourist attractions they visit. Unit testing of this application shows that no errors were found in the application functions. Usability test results were also very positive, with 88% of tourist users stating they were willing to use this application again in the future. This shows that this tour guide application has succeeded in providing an effective solution to the problems faced by tourists in searching for information on facilities at tourist attractions. Thus, this application can be a valuable tool in enhancing the tourist experience and supporting the growth of the Indonesian tourism industry. This research has produced an Android tourist guide application to facilitate travelers in finding public facilities at the "10 New Bali Tourism" destinations. With an 88% user satisfaction rate, the main contribution involves improving the traveler experience, supporting the growth of the tourism industry, and making a positive contribution to the Indonesian economy
Perbandingan Kinerja Algoritma K-means dan Agglomerative Clustering Untuk Segmentasi Penjualan Online Pada Customer Retail
No full textThis research focuses on the comparison between two popular algorithms in data science, namely K-means and Agglomerative Clustering algorithms. The main context of this research is customer data segmentation, a very important process in the business world to understand and serve customers better. The main objective of this research is to evaluate and compare the performance of the two algorithms in generating effective and efficient customer segments. In this research, the dataset used is a retail customer dataset. This dataset includes various attributes that reflect customer characteristics and behavior. To measure the performance of both algorithms, this research uses the RFM (Recency, Frequency, Monetary) weighting method. This method is a commonly used method in customer analysis to identify the most valuable customers based on how recently they transacted (Recency), how often they transact (Frequency), and how much their transactions are worth (Monetary). In addition, this research also uses an evaluation metric known as silhouette score. This metric is used to measure how well an object fits into its own cluster compared to other clusters. The results of this study provide valuable insights into the quality of both algorithms in segmenting customer data. It was found that the K-Means algorithm produced a silhouette score value of 0.5087, while Agglomerative Clustering produced a higher value of 0.6363. This suggests that, in the context of this dataset, Agglomerative Clustering may be more effective compared to K-Means. However, further research is certainly needed to validate these findings and to further explore how these two algorithms can be optimized for customer data segmentatio
Algoritma Principal Component Analysis (PCA) dan Metode Bounding Box pada Pengenalan Citra Wajah
No full textCurrent facial recognition system utilizes the Principal Component Algorithm (PCA) and the Bounding Box method to recognize facial locations based on brightness levels. The problem that was found in the experiment was that unclear or illuminated light factors could cause inaccuracies in facial area recognition. PCA is an algorithm capable of performing dimensional reduction to recognize the face area. The recognition process involves image pre-processing, PCA analysis to produce vectors, and application of a Bounding Box to focus on critical areas. This research contributes to the development of reliable and efficient facial recognition systems, potentially applied in security and access management. The experiment used the Grimace dataset using the .jpg format, with tests on normal brightness and -50 decreases in brightness level. At the decrease in -50, the result shows that the smallest distance value is 3540.1, and the greatest distance is 6849.4 with the average value being 5810.110. face recognition results can recognize face images with the original imag
Teknologi Computer Vision untuk melakukan Deteksi dan Penentuan Kualitas Bibit Ayam Day Old Chicks
No full textThis research is motivated by the importance of chicken animal protein for child growth, development, immunity, intelligence, and the prevention of stunting. This research aims to design and implement an object detection system using a series of 4 nodes, namely Raspberry Pi, ESP32-CAM, Arduino, and ESP32, to identify chickens based on their physical characteristics. This research utilizes computer vision and artificial intelligence methodology, particularly the Convolutional Neural Network approach, to detect characteristics of DOC chickens, such as feathers, eyes, and legs. The implementation results show that the object detection system built using these four nodes can detect objects according to the existing labels. This system can identify DOC chickens with characteristics such as clean feathers, bright eyes, and bright and undamaged legs. Testing was conducted under various movement conditions of the detection objects, and the results show that the system can work well in recognizing the target objects. In the trial section, the objects used were chickens that fit the DOC characteristic category. The trial results show that the built object detection system can detect DOC chickens with suitable physical characteristics. This can assist farmers in selecting and cultivating quality chickens. Five trial tests were conducted, which showed varying detection performance
Optimalisasi Penggunaan Sensor Pada Sistem Penyiraman Tanaman Kangkung Menggunakan Metode WSN
No full textWater spinach is one of the agricultural products that is often consumed by Indonesian people, so it has very high market demand. However, there are still many farmers who use the traditional method of watering plants by carrying water so that it does not increase plant productivity. This is also a problem because farmers are at risk of developing MSDs (musculoskeletal disorders). To support high market demand, technology is needed that can help farmers in the process of watering plants with high effectiveness and efficiency. In this research, optimization of the use of sensors was carried out using WSN (Wireless Sensor Network) to overcome the problem of increasing scale on large areas of land to be able to monitor park land and control the watering process. This system uses several sensors, including a water pressure sensor, waterflow sensor, soil pH sensor and soil moisture sensor. Apart from that, this research also created a warning system and recommendation system to help farmers in making decisions regarding plantation management. Farmers can monitor garden land and control watering from anywhere via the web application that has been create
Teknologi Deteksi Dini Banjir Daerah Aliran Sungai menggunakan Heltec Wifi LoRa 32 V2
No full textIn Indonesia there are often natural disasters, one of which is flooding. Flooding is a natural disaster that is marked by the overflowing of river water irrigation channels in urban areas, one is the river Irrigation that exists at the Technokrat University of Indonesia. Therefore, the study aims to develop a flood early detection tool using LoRa (Long Range) technology to monitor potential flooding in Kalibalau, Indonesian Technocratic University, Bandar Lampung. The research method involves installing an ultrasonic sensor in the Kalibalau River and connecting it to the Heltec Wifi LoRa 32 V2 microcontroller. Test results show that the LoRa transmitter and receiver operate as planned. This tool does not require an internet connection because it uses the Heltec Wifi LoRa 32 V2. The status of the river is categorized into four: Safe, Alert 1, Alert 2, and Danger, with appropriate warnings. The test showed a delay of 5 seconds on the water height reading. At safe (water height 44 cm), the buzzer does not sound. At morning 1 (water altitude 82 cm), it sounds once with a 1 minute delay. The device has a communication capacity of up to 400 meters. Thus, the tool is effective in monitoring the Kalibalau river and giving early warning of potential floods. This research has contributed to the development of flood monitoring technology to increase public alertness and safety in flood-prone area