Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
pada Chatbot-based Information Service using RASA Open-Source Framework in Prambanan Temple Tourism Object
No full textThe pandemic has caused a shift in the tourism industry's drive towards comprehensive digitization. This approach is used to prevent the spread of the Covid-19 virus. The impact of Pemberlakuan Pembatasan Kegiatan Masyarakat (PPKM) limiting the mobility of tourists who will vacation in Indonesia causes losses and foreign exchange earnings of the state in the tourism industry sector of 20.7 billion. So, to survive in the current situation, industry players must be able to adapt and rise by providing more effective innovations. This study aims to develop a Question Answering System or a digital question and answer system using a chatbot (ChatterBot). The chatbot is used as an information service provider that can make it easier for tourists who are looking for information about tourist attractions. Chatbot-based information service systems can work 24 hours or all day, reducing the intensity of direct physical contact with officers and saving operational costs. The chatbot implementation is built on the Machine Learning Framework using RASA Open Source with the Python programming language. The knowledge base of the chatbot system is trained based on the FAQ (Frequently Asking Question) dataset with a case study of the Prambanan Temple tourist attraction as a sample of Indonesian tourism. The results of the evaluation and system performance based on data testing obtained the level of model accuracy is 0.91. Furthermore, the weighted average value in the Confusion Matrix produces a precision of 0.97, a recall of 0.94, and an F1-score of 0.95. The training and testing model processes locally using the Visual Studio Code software.
The pandemic has caused a shift in the tourism industry's drive towards comprehensive digitization. This approach is used to prevent the spread of the Covid-19 virus. The impact of Pemberlakuan Pembatasan Kegiatan Masyarakat (PPKM) limiting the mobility of tourists who will vacation in Indonesia causes losses and foreign exchange earnings of the state in the tourism industry sector of 20.7 billion. So, to survive in the current situation, industry players must be able to adapt and rise by providing more effective innovations. This study aims to develop a Question Answering System or a digital question and answer system using a chatbot (ChatterBot). The chatbot is used as an information service provider that can make it easier for tourists who are looking for information about tourist attractions. Chatbot-based information service systems can work 24 hours or all day, reducing the intensity of direct physical contact with officers and saving operational costs. The chatbot implementation is built on the Machine Learning Framework using RASA Open Source with the Python programming language. The knowledge base of the chatbot system is trained based on the FAQ (Frequently Asking Question) dataset with a case study of the Prambanan Temple tourist attraction as a sample of Indonesian tourism. The results of the evaluation and system performance based on data testing obtained the level of model accuracy is 0.91. Furthermore, the weighted average value in the Confusion Matrix produces a precision of 0.97, a recall of 0.94, and an F1-score of 0.95. The training and testing model processes locally using the Visual Studio Code software
Support Vector Regression Method for Predicting Off-Grid Photovoltaic Output Power in the Short Term
No full textPhotovoltaic (PV) technology is a renewable technology utilizing conversion of solar power or solar radiation into electrical energy. In the manufacture of Solar Power Generation systems, reference is needed regarding the cost of generation and scheduling of maintenance plans. To obtain this reference, it is necessary to predict the photovoltaic power output which is used to determine the power output of PV in the future. In this study, a system that is used to predict short-term power output in PV is designed. This system uses solar irradiation data and 42 days of power output in off-grid PV mini-grid as the dataset. The dataset obtained from the PV output is processed using the Support Vector Regression method with the Kernel Radial Basis Function (RBF) function. Based on the dataset used, this study succeeded in testing the best kernel, namely the RBF kernel. Evaluation of the prediction model obtained a smaller error value than other kernel tests with a Mean Absolute Percentage Error (MAPE) value of 21.082%, Mean Square Error (MSE) value of 0.122, and Mean Absolute Error (MAE) value of 0.262. The prediction model obtained is used to predict the short-term PV power output for the next 3 days. The results of the prediction model have an error value of 5.785 % for MAPE, 0.005 for MAE and 0.069 for MSE. Therefore, the predictive model can be categorized as very good and feasible to predict short-term power outputPhotovoltaic (PV) technology is a renewable technology utilizing conversion of solar power or solar radiation into electrical energy. In the manufacture of Solar Power Generation systems, reference is needed regarding the cost of generation and scheduling of maintenance plans. To obtain this reference, it is necessary to predict the photovoltaic power output which is used to determine the power output of PV in the future. In this study, a system that is used to predict short-term power output in PV is designed. This system uses solar irradiation data and 42 days of power output in off-grid PV mini-grid as the dataset. The dataset obtained from the PV output is processed using the Support Vector Regression method with the Kernel Radial Basis Function (RBF) function. Based on the dataset used, this study succeeded in testing the best kernel, namely the RBF kernel. Evaluation of the prediction model obtained a smaller error value than other kernel tests with a Mean Absolute Percentage Error (MAPE) value of 21.082%, Mean Square Error (MSE) value of 0.122, and Mean Absolute Error (MAE) value of 0.262. The prediction model obtained is used to predict the short-term PV power output for the next 3 days. The results of the prediction model have an error value of 5.785 % for MAPE, 0.005 for MAE and 0.069 for MSE. Therefore, the predictive model can be categorized as very good and feasible to predict short-term power outpu
Depression Detection on Twitter Social Media Using Decision Tree
Get PDFDepression is a major mood illness that causes patients to experience significant symptoms that interfere with their daily activities. As technology has developed, people now frequently express themselves through social media, especially Twitter. Twitter is a social media platform that allows users to post tweets and communicate with each other. Therefore, detecting depression based on social media can help in early treatment for sufferers before further treatment. This study created a system to detect if a person is indicating depression or not based on Depression Anxiety and Stress Scale - 42 (DASS-42) and their tweets using the Classification and Regression Tree (CART) method with TF-IDF feature extraction. The results show that the most optimal model achieved an accuracy score of 81.25% and an f1 score of 85.71%, which are higher than baseline results with an accuracy score of 62.50% and an f1 score of 66.66%. In addition, we found that there were significant effects on changing the value of the maximum features in TF-IDF and changing the maximum depth of the tree to the model performance.Depression is a major mood illness that causes patients to experience significant symptoms that interfere with their daily activities. As technology has developed, people now frequently express themselves through social media, especially Twitter. Twitter is a social media platform that allows users to post tweets and communicate with each other. Therefore, detecting depression based on social media can help in early treatment for sufferers before further treatment. This study created a system to detect if a person is indicating depression or not based on Depression Anxiety and Stress Scale - 42 (DASS-42) and their tweets using the Classification and Regression Tree (CART) method with TF-IDF feature extraction. The results show that the most optimal model achieved an accuracy score of 81.25% and an f1 score of 85.71%, which are higher than baseline results with an accuracy score of 62.50% and an f1 score of 66.66%. In addition, we found that there were significant effects on changing the value of the maximum features in TF-IDF and changing the maximum depth of the tree to the model performance.
 
Buzzer Detection on Indonesian Twitter using SVM and Account Property Feature Extension
Get PDFThe rapid use of Twitter social media in recent times has an impact on the faster dissemination of disinformation which is very dangerous to followers. Detection of disinformation is very important to do and can be done manually by conducting in-depth information analysis. But given the huge amount of information, this approach is less effective. Another, more effective approach is to use a machine learning-based approach. Several studies on hoax information detection based on machine learning have been carried out where some studies analyze the content of a tweet and some others analyze hashtags which are the context of a tweet. The feature usually used to analyze hashtag sentiment data is the property feature of the creator's account. The creator accounts of disinformation are called buzzer accounts. This research proposes account property feature expansion of buzzer accounts combined with the SVM classifier which in several previous similar studies has a very good performance to detect the buzzer hashtag. The experimental results show that expanding the proposed feature can increase SVM's performance in detecting hashtag buzzers by more than 24% compared to using the baseline feature, and the average F1 score obtained from the combination of methods is 84%.
The rapid use of Twitter social media in recent times has an impact on the faster dissemination of disinformation which is very dangerous to followers. Detection of disinformation is very important to do and can be done manually by conducting in-depth information analysis. But given the huge amount of information, this approach is less effective. Another, more effective approach is to use a machine learning-based approach. Several studies on hoax information detection based on machine learning have been carried out where some studies analyze the content of a tweet and some others analyze hashtags which are the context of a tweet. The feature usually used to analyze hashtag sentiment data is the property feature of the creator's account. The creator accounts of disinformation are called buzzer accounts. This research proposes account property feature expansion of buzzer accounts combined with the SVM classifier which in several previous similar studies has a very good performance to detect the buzzer hashtag. The experimental results show that expanding the proposed feature can increase SVM's performance in detecting hashtag buzzers by more than 24% compared to using the baseline feature, and the average F1 score obtained from the combination of methods is 84%
Deteksi Penyakit Covid-19 Pada Citra X-Ray Dengan Pendekatan Convolutional Neural Network (CNN)
Get PDFThe Coronavirus (COVID-19) pandemic has resulted in the worldwide death rate continuing to increase significantly, identification using medical imaging such as X-rays and computed tomography plays an important role in helping medical personnel diagnose positive negative COVID-19 patients, several works have proven the learning approach in-depth using a Convolutional Neural Network (CNN) produces good accuracy for COVID detection based on chest X-Ray images, in this study we propose different transfer learning architectures VGG19, MobileNetV2, InceptionResNetV2 and ResNet (ResNet101V2, ResNet152V2 and ResNet50V2) to analyze their performance, testing conducted in the Google Colab work environment as a platform for creating Python-based applications and all datasets are stored on the Google Drive application, the preprocessing stages are carried out before training and testing, the datasets are grouped into theNormal and COVID folders then combined m become a set of data by dividing them into training sets of 352 images, testing 110 images and validating 88 images, then the detection results are labeled with the number 1 means COVID and the number 0 for NORMAL. Based on the test results, the ResNet50V2 model has a better accuracy rate than other models with an accuracy level of about 0.95 (95%) Precision 0.96, Recall 0.973, F1-Score 0.966, and Support of 74, then InceptionResNetV2, VGG19, and MobileNetV2, so that ResNet50V2-based CNNs can be used as initial identification for the classification of a patientinfected with COVID or NORMAL.Pandemi Coronavirus (COVID-19) telah mengakibatkan tingkat kematian di seluruh dunia terus meningkat secara signifikan, identifikasi menggunakan pencitraan medis seperti: Sinar-X dan computed tomography memainkan peran penting dalam membantu medis mendiagnosis pasien COVID-19 positif atau negatif, beberapa penelitian telah membuktikan pendekatan pembelajaran secara mendalam menggunakan Convolutional Neural Network (CNN) menghasilkan akurasi yang baik untuk deteksi COVID berdasarkan rontgen dada gambar, dalam penelitian ini kami mengusulkan arsitektur pembelajaran transfer yang berbeda VGG19, MobileNetV2, InceptionResNetV2 dan ResNet (ResNet101V2, ResNet152V2 dan ResNet50V2) untuk menganalisis kinerjanya, pengujian dilakukan di Google Lingkungan kerja Colab sebagai platform untuk membuat aplikasi berbasis Python dan semua dataset disimpan di aplikasi Google Drive, pra-pemrosesan tahapan yang dilakukan sebelum pelatihan dan pengujian, dataset dikelompokkan kedalam Folder Normal dan COVID kemudian digabungkan menjadi satu dataset dengan membagi kedalam pelatihan 352 gambar, menguji 110 gambar dan memvalidasi 88 gambar, kemudian hasil deteksi diberi label dengan angka 1 artinya COVID dan angka 0 untuk NORMAL. Berdasarkan hasil pengujian, ResNet50V2 model memiliki tingkat akurasi yang lebih baik dibandingkan model lain dengan tingkat akurasi sekitar 0,95 (95%) Presisi 0,96, Recall 0,973, F1-Score 0,966, dan Dukungan 74, lalu InceptionResNetV2, VGG19, dan MobileNetV2, sehingga ResNet50V2- CNN berbasis dapat digunakan sebagai identifikasi awal untuk klasifikasi pasien terinfeksi COVID atau NORMAL
Feature Expansion Word2Vec Untuk Analisis Sentimen Kebijakan Publik di Twitter
No full textSocial media users, especially on Twitter, can freely express opinions or other information in the form of tweets about anything, including responding to a public policy. In a written tweet, there is a limit of 280 characters per tweet and this allows for problems such as vocabulary mismatches. Therefore, in this study, the feature expansion Word2vec method was applied to overcome when the vocabulary mismatches occur. This study develops and compares the Twitter sentiment analysis system using the feature expansion Word2vec method with the Logistic Regression (LR) and Support Vector Machine (SVM) classification algorithms and the system without the feature expansion Word2Vec method. The results of this study, the feature expansion Word2Vec method on the SVM classification algorithm succeeded in increasing the system accuracy up to 0,99% with an accuracy value of 78,99%
Prediksi Harga Cryptocurrency Menggunakan Algoritma Long Short Term Memory (LSTM)
No full textTechnological developments continue to encourage the creation of various innovations in almost all aspects of human life. One of the innovations that is becoming a worldwide phenomenon today is the presence of cryptocurrency as a digital currency that is able to replace the role of conventional currency as a means of payment. Currently, the number of cryptocurrency investors in Indonesia has reached 4.45 million people as of March 2021, an increase of 78% compared to the end of the previous year. Very volatile price movements make cryptocurrency investments considered speculative so the risks faced are also very high. The purpose of this study is to build a predictive model that is able to forecast prices on the cryptocurrency market. The algorithm used to build the prediction model is Long Short Term Memory (LSTM). LSTM is the development of the Recurrent Neural Network (RNN) algorithm to overcome problems in the RNN in managing data for a long period. LSTM is considered superior to other algorithms in managing time series data. The data in this study were taken from the Yahoo Finance website using the Pandas Datareader library through Google Collaboratory. The entire prediction model development process is carried out through Google Collaboratory tools. To improve the accuracy of the model, the Nadam optimization algorithm was used and three testing sessions were carried out with the number of Epochs of 1, 10, and 20 in each session. The final test results show that the best prediction performance occurs when testing the DOGE coin type with the number of Epoch 20 which gets an RMSE value of 0.0630.Perkembangan teknologi terus mendorong terciptanya berbagai inovasi hampir diseluruh aspek kehidupan manusia. Salah satu inovasi yang menjadi fenomena di seluruh dunia saat ini adalah hadirnya cryptocurrency sebagai mata uang digital yang mampu menggantikan peran mata uang konvensional sebagai alat pembayaran. Saat ini jumlah investor cryptocurrency di Indonesia telah mencapai angka 4,45 juta orang per bulan Maret 2021 mengalami peningkatan 78% dibandingkan akhir tahun sebelumnya. Pergerakan harga yang sangat fluktuatif menjadikan investasi cryptocurrency dianggap spekulatif sehingga risiko yang dihadapi juga sangat tinggi. Tujuan penelitian ini adalah untuk membangun model prediksi yang mampu melakukan peramalan harga pada pasar cryptocurrency. Algoritma yang digunakan untuk membangun model prediksi adalah Long Short Term Memory (LSTM). LSTM merupakan pengembangan dari algoritma Recurrent Neural Network (RNN) untuk mengatasi permasalahan pada RNN dalam mengelola data untuk periode yang lama. LSTM dianggap lebih unggul dibandingakan algoritma lainnya dalam mengelola data yang bersifat time series. Data pada penelitian ini diambil dari situs Yahoo Finance dengan menggunakan library Pandas Datareader melalui Google Colaboratory. Keseluruhan proses pembangunan model prediksi dilakukan melalui tools Google Colaboratory. Untuk meningkatkan akurasi model digunakan algoritma optimasi Nadam serta dilakukan tiga sesi pengujian dengan jumlah Epoch masing-masing 1, 10, dan 20 pada tiap sesi. Hasil akhir pengujian menunjukan performa prediksi terbaik terjadi pada saat melakukan pengujian terhadap jenis koin DOGE dengan jumlah Epoch 20 yang mendapatkan nilai RMSE sebesar 0,0630
Design of Smart Farm Irrigation Monitoring System Using IoT and LoRA
Get PDFAgriculture is an essential part of society in Indonesia because most of the population lives off of farming. Water and irrigation are the most critical and central factors in the agricultural system. The uneven distribution of irrigation water can be a problem for farmers. In addition, most of the current irrigation systems are still operated manually, for example, irrigation gates. The gate still works manually and requires human labor to run it. This study aims to design a smart farm irrigation system using internet of things and LoRa communication technology. LoRa can transmit information up to a range of several kilometers without an internet connection. It will be advantageous when the farm's location is deep in the forest, and there is no GSM signal for internet access. This study indicates that this system brings benefits for farmers in running agriculture. Farmers' work time is shorter. They can use the remaining time to do other businesses to increase their income. The irrigation monitoring process becomes easier because they do not need to come to the farm location. In fact, they can use smartphones to monitor it.  
A Comparison of Deep Learning Approach for Underwater Object Detection
Get PDFIn recent years, marine ecosystems and fisheries have become potential resources. Therefore, monitoring these objects will be essential to ensure their existence. One of the computer vision techniques is object detection, utilized to recognize and localize objects in underwater scenery. Many studies have been conducted to investigate various deep learning methods implemented in underwater object detection; however, only a few investigations have been performed to compare mainstream object detection algorithms in these circumstances. This article examines various state-of-the-art deep learning methods applied to underwater object detection, including Faster-RCNN, SSD, RetinaNet, YOLOv3, and YOLOv4. We trained five models on the RUIE dataset. The average detection time was used to compare how fast a model can detect an object within an image, and mAP was also applied to measure detection accuracy. All trained models have costs and benefits; SSD was fast but had poor performance; RetinaNet had consistent performance across different thresholds, but the detection speed was slow; YOLOv3 was the fastest and had acceptable performance comparable with RetinaNet; YOLOv4 was good at first, but performance dropped as threshold enlargement; also, YOLOv4 needed extra time to detect objects compared to YOLOv3. There are no models that are fully suited for underwater object detection; nonetheless, when the mAP and average detection time of the five models were compared, we determined that YOLOv3 is the best acceptable model among the evaluated underwater object detection models.
 
Classification of Face Mask Detection Using Transfer Learning Model DenseNet169
Get PDFCOVID-19 has become a threat to the world because it has spread throughout the world. The fight against this pandemic is becoming an unavoidable reality for many countries. The government has set policies on various transmission prevention efforts. One of these efforts is for everyone to wear masks to break the transmission chain. With such conditions, the government must continue to monitor so that people can apply the appeal in their daily lives when participating in outdoor activities. The present time involves new problems in so many fields of information technology research, especially those related to artificial intelligence. The purpose of this study is to discuss the classification of face image detection in people who wear masks and do not wear masks. designed using the Convolutional Neural Network (CNN) model and built using the transfer learning method with the DenseNet169 model. The model used is also combined with the DenseNet169 transfer learning method and the fully connected layer model architecture, to optimize the performance test in the evaluation. These models were trained under similar conditions and evaluated on benchmarks with the same training and validation images. The result of this research is to get an accuracy value of 96% by combining the two datasets. This dataset is the same as previous research; the number of datasets is 8929 imagesCOVID-19 has become a threat to the world because it has spread throughout the world. The fight against this pandemic is becoming an unavoidable reality for many countries. The government has set policies on various transmission prevention efforts. One of these efforts is for everyone to wear masks in order to break the transmission chain. With such conditions, the government must continue to monitor so that people can apply the appeal in their daily lives when participating in outdoor activities. The present time involves new problems in so many fields of information technology research, especially those related to artificial intelligence. The purpose of this study is to discuss the classification of face image detection in people who wear masks and do not wear masks. designed using the Convolutional Neural Network (CNN) model and built using the transfer learning method with the DenseNet169 model. The model used is also combined with the DenseNet169 transfer learning method and the fully connected layer model architecture, so as to optimize the performance test in the evaluation. These models were trained under similar conditions and evaluated on benchmarks with the same training and validation images. The result of this research is to get an accuracy value of 96% by combining the two datasets. This dataset is the same as previous research; the number of datasets is 8929 image