Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
Using Histogram Extracted From Satelite Imagery and Convolutional Network to Predict GRDP in Java Region
Get PDFInequality is one of the problems faced by all countries in the world, including Indonesia. The data used to measure development inequality between regions mostly use GRDP data. However, the GRDP data issued by BPS has a deficiency, it was released after the current year, and this figure is provisional. Therefore, a new data source is needed that can be used to estimate the value of economic activity so that it can be used to measure the level of inequality in development in a region. Nighttime Light (NTL) satellite imagery data can be an alternative to see socioeconomic activity in an area and have been shown to have a strong correlation with socioeconomic activity. In this study, we used VIIRS NTL satellite imagery data and Dynamic World land cover data to estimate GRDP. Rather than using statistical features for each area of interest, we use features in the form of histograms extracted from NTL images and land cover images for each area of interest. Using a histogram, we do not lose spatial information from satellite imagery. Then we proposed a deep learning method in the form of a one-dimensional convolutional neural network using the Huber loss function. This model obtained good precision with an R-square value of 0.8549, beating the baseline method with two-dimensional convolutional networks. The use of the Huber loss function can improve the performance of the model, which has a smaller total loss and has a smoother gradient
Comparison of LSTM and IndoBERT Method in Identifying Hoax on Twitter
Get PDFIn recent years, social media users have been increasing significantly, in January 2022 social media users in Indonesia reached 191 million people which has an increase of 12.35% from the previous year as many as 170 million people, With this massive increase every year, more and more people tend to seek and consume information through social media. Despite the many advantages provided by social media, However, the quality of information on social media is lower than in traditional news media there is a lot of hoax information spreading. With many disadvantages felt by hoax information, it has led to many research to detect hoax information on social media, especially information that is widely spread on Twitter. There are several previous researches that use various models using machine learning and also using deep learning to detect hoax. deep learning is very well used to perform several text classification tasks, especially in detecting hoax. The aim of this paper is to compare the LSTM and IndoBERT methods in detecting hoax using datasets taken from Twitter. In this study, two experiments work are conducted, LSTM and IndoBERT methods. The experimental results is average value obtained from experiments using 10-fold cross-validation. The IndoBERT model shows good performance with an average accuracy value of 92.07%, and the LSTM model provides an average accuracy value of 87.54%. The IndoBERT model can show good performance in hoax detection tasks and is shown to outperform the LSTM model which can provide the best average accuracy results in this study.In recent years, social media users have been increasing significantly, in January 2022 social media users in Indonesia reached 191 million people which has an increase of 12.35% from the previous year as many as 170 million people, With this massive increase every year, more and more people tend to seek and consume information through social media. Despite the many advantages provided by social media, However, the quality of information on social media is lower than in traditional news media there is a lot of hoax information spreading. With many disadvantages felt by hoax information, it has led to many research to detect hoax information on social media, especially information that is widely spread on Twitter. There are several previous researches that use various models using machine learning and also using deep learning to detect hoax. deep learning is very well used to perform several text classification tasks, especially in detecting hoax. The aim of this paper is to compare the LSTM and IndoBERT methods in detecting hoax using datasets taken from Twitter. In this study, two experiments work are conducted, LSTM and IndoBERT methods. The experimental results is average value obtained from experiments using 10-fold cross-validation. The IndoBERT model shows good performance with an average accuracy value of 92.07%, and the LSTM model provides an average accuracy value of 87.54%. The IndoBERT model can show good performance in hoax detection tasks and is shown to outperform the LSTM model which can provide the best average accuracy results in this study
Clustering Analysis and Mapping of ISPA Disease Spread Patterns in Bireuen District
Get PDFISPA disease can be detected by analyzing the regional distribution map of the disease. Early detection of ARI is very important for effective prevention. The study conducted in Bireuen Regency used data from 2019 to 2021, sourced from dr. Fauziah Bireuen Hospital, revealed that there was an increase in ARI cases from an average of 13.18 to 59.24 per year. The aims of the study were to identify ARI clusters, analyze disease patterns using Spatial Pattern Analysis and Flexible Shaped Spatial Scanning Statistics. The methodology involves collecting patient data for each ARI case and processing it using DBSCAN to obtain cluster points on the map. Spatial Pattern Analysis is used to analyze these clusters and identify hotspot points on the map. The analysis resulted in four clusters: Cluster 1 (6 subdistrict), Cluster 2 (4 subdistrict), Cluster 3 (1 subdistrict), and Cluster 4 (6 subdistrict). The study identified 6 hotspots in 2019, 5 hotspots in 2020, and 6 hotspots in 2021. Each ARI disease clustering map shows the distribution of ARI cases and identifies areas prone to the disease. These findings provide valuable insights for targeted interventions and preventive actions in identified high-risk areas of ISPA
Comparison of the RFM Model's Actual Value and Score Value for Clustering
Get PDFClustering algorithms and Recency-Frequency-Money (RFM) models are widely implemented in various sectors of e-commerce, banking, telecommunications and other industries to obtain customer segmentation. The RFM model will assess a line of data which includes the recency and frequency of data appearance, as well as the monetary value of a transaction made by a customer. Choosing the right RFM model also influences the analysis of cluster results, the output of cluster results is more compact for the same clusters (inter-cluster) and separate for other clusters (intra-cluster). Through an experimental approach, this research aims to find the best data set transformation model between actual RFM values and RFM scores. The method used is to compare the actual RFM value model and the RFM score and use the silhouette score value as an indicator to obtain the best clustering results using the K-Means algorithm. The subject of this research is a stall-based e-Commerce application, where data was taken in the Wiradesa area, Central Java. The resulting data set consisted of 273,454 rows with 18 attributes from January 2022 to December 2022 by collecting historical data from shopping outlets to wholesalers. The analysis of the data set was carried out by transforming the data set using the RFM method into actual values and score values; then the dataset was used to obtain the best cluster data. The results of this research show that transaction data based on time (time series) can be transformed into data in the RFM model where the actual value is better than the RFM score model with a silhouette score = 0.624646 and the number of clusters (K) =3. The results of the clustering process also form a series of data with a cluster label, thus forming supervised learning data
A Hybrid Method on Emotion Detection for Indonesian Tweets of COVID-19
Get PDFAs a result of the COVID-19 pandemic, there have been restrictions on activities outside the home which has caused people to interact more and express their emotions through social media platforms, one of which is Twitter. Previous studies on emotion classification used only one feature extraction, namely the lexicon based or word embedding. Feature extraction using the emotion lexicon has the advantage of recognizing emotional words in a sentence while feature extraction using word embedding has the advantage of recognizing the semantic meaning. Therefore, the main contribution to this research is to use two lexicon feature extraction and word embedding to classify emotions. The classification technique used in this research is the Ensemble Voting Classifier by selecting the two best classifiers to try on both types of feature extraction. The experimental results for both types of feature extraction are the same, indicating that the best classifiers are Random Forest and SVM. Models using both types of feature extraction show increased accuracy compared to using only one feature extraction. The results of this emotional analysis can be used to determine the public's reaction to an event, product, or public policy.As a result of the COVID-19 pandemic, there have been restrictions on activities outside the home which has caused people to interact more and express their emotions through social media platforms, one of which is Twitter. Previous studies on emotion classification used only one feature extraction, namely the lexicon based or word embedding. Feature extraction using the emotion lexicon has the advantage of recognizing emotional words in a sentence while feature extraction using word embedding has the advantage of recognizing the semantic meaning. Therefore, the main contribution to this research is to use two lexicon feature extraction and word embedding to classify emotions. The classification technique used in this research is the Ensemble Voting Classifier by selecting the two best classifiers to try on both types of feature extraction. The experimental results for both types of feature extraction are the same, indicating that the best classifiers are Random Forest and SVM. Models using both types of feature extraction show increased accuracy compared to using only one feature extraction. The results of this emotional analysis can be used to determine the public's reaction to an event, product, or public policy
Solution to Scalability and Sparsity Problems in Collaborative Filtering using K-Means Clustering and Weight Point Rank (WP-Rank
Get PDFCollaborative filtering is a method that can be used in recommendation systems. Collaborative Filtering works by analyzing rating data patterns. It is also used to make predictions of interest to users. This process begins with collecting data and analyzing large amounts of information on the behavior, activities, and tendencies of users. The results of the analysis are used to predict what users like based on similarities with other users. In addition, collaborative filtering is able to produce recommendations of better quality than recommendation systems based on content and demographics. However, collaborative filtering still faces scalability and sparsity problems. It are because the data is always evolving so that it becomes big data, besides that there are many data with incomplete conditions or many vacancies are found. Therefore, the purpose of this study proposed a clustering and ranking-based approach. The cluster algorithm used K-Means. Meanwhile, the WP-Rank method was used for ranking based. The experimental results showed that the running time was faster with an average execution time of 0.15 seconds by clustering. Furthermore, it was able to improve the quality of the recommendations, as indicated by an increase in the value of NDCG at k=22, the average value of NDCG was 0.82, so the recommendations produced were higher quality and more appropriate to the interests of the users.
Collaborative Filtering is a method to be used in recommendation systems. Collaborative Filtering works by analyzing rating data patterns. It is also used to make predictions of user interest. This process begins with collecting data and analyzing large amounts of information about the behavior, activities, and tendencies of users. The results of the analysis are used to predict what users like based on similarities with other users. In addition, Collaborative Filtering is able to produce recommendations with better quality than recommendation systems based on content and demographics. However, Collaborative Filtering still faces scalability and sparsity problems. It is because the data is always evolving so that it becomes big data, besides that there are many data with incomplete conditions or many vacancies are found. Therefore, the purpose of this study proposed a clustering and ranking based approach. The cluster algorithm used K-Means. Meanwhile, the WP-Rank method was used for ranking based. The experimental results showed that the running time was faster with an average execution time of 0.15 second by clustering. In addition, it was able to improve the quality of recommendations as indicated by an increase in the value of NDCG at k=22, the average value of NDCG was 0.82, so that the recommendations produced had more quality and more appropriate with user interests.
 
Multi-Accent Speaker Detection Using Normalize Feature MFCC Neural Network Method
Get PDFSpeaker recognition is a field of research that continues to this day. Various methods have been developed to detect the human voice with greater precision and accuracy. Research on human speech recognition that is quite challenging is accent recognition. Detecting various types of human accents with different accents and ethnicities with high accuracy is a research that is quite difficult to do. According to the results of the research on the data preprocessing stage, feature extraction and selection of the right classification method play a very important role in determining the accuracy results. This study uses a preprocessing approach with normalizing features combined with MFCC as a method to perform feature extraction and the neural network (NN), which is a classification method that works based on the workings of the human brain. Research results obtained using the normalize feature with MFCC and neural network for multiaccent speaker recognition, the accuracy performance reaches 82.68%, precision is 83% and recall is 82.88%.
Speaker recognition is a field of research that continues to this day. Various methods have been developed to detect the human voice with greater precision and accuracy. Research on human speech recognition that is quite challenging is accent recognition. Detecting various types of human accents with different accents and ethnicities with high accuracy is a research that is quite difficult to do. According to the results of the research on the data preprocessing stage, feature extraction and the selection of the right classification method play a very important role in determining the accuracy results. This study uses a preprocessing approach with normalizing features combined with MFCC as a method for performing feature extraction and Neural Network (NN) which is a classification method that works based on the workings of the human brain. Research results obtained using the normalize feature with MFCC and Neural Network for multi-accent speaker recognition, the accuracy performance reaches 82.68%, precision is 83% and recall is 82.88%
Classification of Hearing Loss Degrees with Naive Bayes Algorithm
Get PDFAccording to the World Health Organization (WHO), hearing loss is one of the fourth leading causes of disability. The number of people with hearing loss continues to increase yearly. This increase occurred due to delays in recognizing hearing loss, leading to delays in providing treatment. To solve this problem, one solution to deal with this is early identification to detect the degree of hearing loss. This research will use machine learning to classify the degree of hearing loss. The algorithm implemented in this study is naive Bayes. This study uses a data set from the Zenodo open access repository with 3105 raw data and 19 features. This study evaluates the performance of overall accuracy, precision, recall, and f1-score and classified four classes: mild, moderate, moderately severe, and severe. The methodology classification stages in this study include data preprocessing, data training, data testing, and evaluation. From evaluating the performance of the Naive Bayes algorithm, the classification results obtained the highest impacts in the form of 94% overall accuracy, 100% precision, 100% recall and 97% f1-score in classifying the degree of hearing loss.
According to the World Health Organization (WHO), hearing loss is one of the fourth highest causes of disability. The number of people with hearing loss continues to increase yearly. This increase occurred due to delays in recognizing the hearing loss, leading to delays in providing treatment. To solve this problem one solution to deal with this is early identification to detect the degree of hearing loss. This research will use machine learning to classify the degree of hearing loss. The algorithm implemented in this study is naive Bayes. This study uses a dataset from the open-access repository Zenodo with 3105 raw data and 19 features. This study evaluates the performance of overall accuracy, precision, recall, and f1-score and classified four classes: mild, moderate, moderately severe, and severe. The methodology classification stages in this study include data pre-processing, data training, data testing to evaluation. From evaluating the performance of the Naive Bayes algorithm, the classification results obtained the highest impacts in the form of 94% overall accuracy, 100% precision, 100% recall, and 97% f1-score in classifying the degree of hearing loss
Implementation of System Development Life Cycle (SDLC) on IoT-Based Lending Locker Application
Get PDFLibraries are social institutions that provide information services that can be accessed publicly to meet the information needs of librarians. Based on the results of a survey conducted on 58 students who have visited the library of the Electrical and Computer Engineering Department of the Universitas Syiah Kuala, an information system was needed that provided research book information related to the author's name, year of writing, field concentration, and abstract of the research book, and there was a division of categories based on field concentration, and there was an online borrowing feature. Based on these problems, this study aims to implement an Android application system with IoT-based lending lockers using the SDLC (system development life cycle) prototyping method. This study produces an application with a locker-based online lending feature, several other features as the user desires, and one prototype lending locker. The locker-based online lending system integrated with ESP32-WROOM-32 can connect to Firebase storage and send locker key codes to Firebase so that the application can access them. Through experiments and tests conducted on the application, it is obtained that the application can access the locker key code and display it to the user. The application has also been validated using black-box and white-box testing and can be accepted by users based on the System Usability Scale (SUS) average score with a very feasible interpretation category.Libraries are social institutions that provide information services that can be accessed publicly to meet the information needs of librarians. Based on the results of a survey conducted on 58 students who have visited the library of the Electrical and Computer Engineering Department of Universitas Syiah Kuala, an information system was needed that provided research book information related to the author's name, year of writing, field concentration, and abstract of the research book and there was a division of categories based on field concentration, and there was an online borrowing feature. Based on these problems, this study aims to implement an Android application system with IoT-based lending lockers using the SDLC (system development life cycle) Prototyping method. This study produces an application with a locker-based online lending feature, several other features as the user desires, and 1 prototype lending locker. The locker-based online lending system integrated with ESP32-WROOM-32 can connect with Firebase storage and send locker key codes to Firebase so the application can access them. Through experiments and tests conducted on the application, it is obtained that the application can access the locker key code and display it to the user. The application has also been validated using black-box and white-box testing and can be accepted by users based on System Usability Scale (SUS) average score with a very feasible interpretation categor
Application of Object Mask Detection Using the Convolution Neural Network (CNN)
No full textThe spread of Coronavirus Disease (Covid-19) is still a serious problem that we are currently facing. Spread occurred very quickly through the face-to-face interaction process. The face-to-face interaction process that occurs both in public spaces and in closed spaces has a great risk of transmitting the Covid-19 virus. One of the efforts to deal with the spread of the Covid-19 virus is to increase the use of masks in both public and closed spaces. On the basis of this, this study aims to develop an object detection process in image processing techniques. Object detection development using the convolution neural network (CNN) method to provide optimal output. CNN can process the input image, which is converted into a pixel matrix and then sent to the convolution layer. The research data set consists of 2000 images of masks and not masks. The images were obtained from open sources, github.com and kaggle.com. The results of the study present a system capable of detecting masks in real time. CNN provides very good performance with an accuracy rate of 99.05%. With these results, the contribution of this research can be used in the monitoring of public services for the community to increase the use of masks.
The spread of Coronavirus Disease (Covid-19) is still a serious problem we are currently facing. The spread occurred very quickly through the process of face-to-face interaction. The process of face-to-face interaction that occurs both in public spaces and closed spaces has a great risk of transmitting the Covid-19 virus. One of the efforts to deal with the spread of the Covid-19 virus is by increasing the use of masks both in public and closed spaces. Based on this, this study aims to develop an Object Detection process in image processing techniques. Object Detection development using the Convolution Neural Network (CNN) method to provide optimal output. The CNN can process the input image which is converted into a pixel matrix and then forwarded to the convolution layer. The research dataset consists of 2000 images of face masks and not masks. The images were obtained from the open sources github.com and kaggle.com. The results of the study present a system capable of detecting masks in real time. CNN provides very good performance with an accuracy rate of 99.05%. With these results, the contribution of this research can be used for the process of monitoring public services for the community to increase the use of masks