Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862853
Nursilva Aulianisa Putri, Esmeralda Contessa Djamal, Fikri Nugraha, Fatan Kasyidi
Electroencephalogram (EEG) signals of certain emotions contain waves with specific frequency bands. So, emotion recognition uses the network containing each wave to become relevant. EEG signals record electrical activity in the brain from several channels. Therefore, EEG signal processing needs consideration to spatial and temporal. Spatial is a signal between channels, while temporal is a sequence. Several methods were used, Convolutional Neural Networks (CNN) with various dimensions, Recurrent Neural Networks (RNN), and hybrid CNN-RNN. This paper proposed a hybrid 2D CNN-RNN method for identifying emotions from a parallel network of each wave. Two-dimensional CNN is used in channel extraction in a short time of the signal. Using short-time signals is intended to minimize the non-stationary characteristic of EEG signals. Meanwhile, the identification of emotions is carried out with RNN using the output of 2D CNN extraction. The modeling and testing used a dataset from SEED, with three emotion classes: positive, neutral, and negative. The experimental results show that using a split network of each wave increased accuracy from 80.92% to 84.71% and a decreased Loss value. While the use of 2D CNN only increased a less significant accuracy than 1D CNN. Evaluation of the waves shows that Beta and Gamma waves provided the best precision, 87-91%, and Theta waves gave 79-85% precision. Alpha wave degrades overall performance, which only has 56-61% precision, considering it is a mid-wave between Theta and Beta. It is necessary to choose the proper weight updating technique. Adaptive Moment (Adam) increased accuracy than AdaDelta, AdaGrad, and RMSprop.
{"title":"EEG Emotion Recognition using Parallel Hybrid Convolutional-Recurrent Neural Networks","authors":"Nursilva Aulianisa Putri, Esmeralda Contessa Djamal, Fikri Nugraha, Fatan Kasyidi","doi":"10.1109/ICoDSA55874.2022.9862853","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862853","url":null,"abstract":"Electroencephalogram (EEG) signals of certain emotions contain waves with specific frequency bands. So, emotion recognition uses the network containing each wave to become relevant. EEG signals record electrical activity in the brain from several channels. Therefore, EEG signal processing needs consideration to spatial and temporal. Spatial is a signal between channels, while temporal is a sequence. Several methods were used, Convolutional Neural Networks (CNN) with various dimensions, Recurrent Neural Networks (RNN), and hybrid CNN-RNN. This paper proposed a hybrid 2D CNN-RNN method for identifying emotions from a parallel network of each wave. Two-dimensional CNN is used in channel extraction in a short time of the signal. Using short-time signals is intended to minimize the non-stationary characteristic of EEG signals. Meanwhile, the identification of emotions is carried out with RNN using the output of 2D CNN extraction. The modeling and testing used a dataset from SEED, with three emotion classes: positive, neutral, and negative. The experimental results show that using a split network of each wave increased accuracy from 80.92% to 84.71% and a decreased Loss value. While the use of 2D CNN only increased a less significant accuracy than 1D CNN. Evaluation of the waves shows that Beta and Gamma waves provided the best precision, 87-91%, and Theta waves gave 79-85% precision. Alpha wave degrades overall performance, which only has 56-61% precision, considering it is a mid-wave between Theta and Beta. It is necessary to choose the proper weight updating technique. Adaptive Moment (Adam) increased accuracy than AdaDelta, AdaGrad, and RMSprop.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123951911","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862907
Dhuhita Trias Maulidia, Erwin Budi Setiawan
Online Social Networks have an essential role as a source of information, especially during an emergency. One of them is Twitter, a service that allows users to send and read messages but is limited in character. Thus, tweets that are written are very short and do not always use the correct grammar and use many variations of words. Using word variations can increase the likelihood of vocabulary mismatches and make tweets difficult to understand. One solution to overcome this problem is to expand the features of the tweet. The feature expansion on Twitter is a semantic addition to the process of multiplying the original text to make it look like large text. In this study, Word2Vec will be used with the Gradient Boosted Decision Tree Method to classify it. The expected result of this research is to reduce words or sentences in the classification of Twitter topics which are evaluated using the accuracy value, F1-Measure. The highest accuracy value in the application of feature expansion using Word2Vec with the Gradient Boosted Decision Tree classification method is 85.44%.
{"title":"Feature Expansion with Word2Vec for Topic Classification with Gradient Boosted Decision Tree on Twitter","authors":"Dhuhita Trias Maulidia, Erwin Budi Setiawan","doi":"10.1109/ICoDSA55874.2022.9862907","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862907","url":null,"abstract":"Online Social Networks have an essential role as a source of information, especially during an emergency. One of them is Twitter, a service that allows users to send and read messages but is limited in character. Thus, tweets that are written are very short and do not always use the correct grammar and use many variations of words. Using word variations can increase the likelihood of vocabulary mismatches and make tweets difficult to understand. One solution to overcome this problem is to expand the features of the tweet. The feature expansion on Twitter is a semantic addition to the process of multiplying the original text to make it look like large text. In this study, Word2Vec will be used with the Gradient Boosted Decision Tree Method to classify it. The expected result of this research is to reduce words or sentences in the classification of Twitter topics which are evaluated using the accuracy value, F1-Measure. The highest accuracy value in the application of feature expansion using Word2Vec with the Gradient Boosted Decision Tree classification method is 85.44%.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115804933","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862822
S. Noviaristanti, G. Ramantoko, Akas Triono Hadi, Alfi Inayati
A private university must consider many things in accepting prospective students. Students enrolled are expected to stay until their studies are completed, have good academic performance, and be able to graduate on time. Private universities, from the beginning of the admission of new students, it is necessary to choose which prospective students are accepted to achieve the quality of education goals in the study program. This work aims to study the prediction class and class order of variable importance to students’ length of stay and academic performance labeled graduation. The method adopted falls into a technique called feature extraction. This study uses rank methods information gain and gain ratio to confront other methods χ2 and random forest. A dataset of 7676 observations, spanning the years from 2010-2021, students from a management program of a private university in Indonesia, is used. This study collects data from the faculty-specific department from the university’s academic admissions as inputs. The result of the study shows that all techniques vote IP/GPA (IP) as the most critical feature in predicting length of stay and graduation. Origin of High School, Selection Test Score, and Gender get split votes. This study is unique because it sheds light on the case particularity to Indonesia.
{"title":"Predictive Model of Student Academic Performance in Private Higher Education Institution (Case in Undergraduate Management Program)","authors":"S. Noviaristanti, G. Ramantoko, Akas Triono Hadi, Alfi Inayati","doi":"10.1109/ICoDSA55874.2022.9862822","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862822","url":null,"abstract":"A private university must consider many things in accepting prospective students. Students enrolled are expected to stay until their studies are completed, have good academic performance, and be able to graduate on time. Private universities, from the beginning of the admission of new students, it is necessary to choose which prospective students are accepted to achieve the quality of education goals in the study program. This work aims to study the prediction class and class order of variable importance to students’ length of stay and academic performance labeled graduation. The method adopted falls into a technique called feature extraction. This study uses rank methods information gain and gain ratio to confront other methods χ2 and random forest. A dataset of 7676 observations, spanning the years from 2010-2021, students from a management program of a private university in Indonesia, is used. This study collects data from the faculty-specific department from the university’s academic admissions as inputs. The result of the study shows that all techniques vote IP/GPA (IP) as the most critical feature in predicting length of stay and graduation. Origin of High School, Selection Test Score, and Gender get split votes. This study is unique because it sheds light on the case particularity to Indonesia.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114665097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Meat is a source of food that contains many nutrients. The nutritional content of meat consists of fat, calories, trans fat, saturated fat, calcium, protein, vitamin D, vitamin B6, vitamin B12, and magnesium. Due to its good nutritional content, the demand for meat in Indonesia has increased. However, there are problems with meat health. Meat is prone to spoilage and is quickly contaminated with microbes. The microbial population can spoil or spoil the meat. Checking the feasibility of meat is usually done by looking at the texture of the meat traditionally. However, this method is less effective in assessing the feasibility of meat. Therefore, another method is used to determine the feasibility of meat, namely using the Electronic Nose (e-nose) with the Neural Network (NN) algorithm. Because by using an e-nose, that can find out the smell or smell of decent meat. They are applying the NN algorithm for classification to work in a structured manner on each component needed to determine meat quality. These results can help people to get the meat of good quality. The experiment was carried out using a dataset that had a total of 2220 data. The experimental results show that using the NN algorithm with the e-nose sensor gets an accuracy of 0.92.
{"title":"Electronic Nose and Neural Network Algorithm for Multiclass Classification of Meat Quality","authors":"Alif Firman Juannata, Dedy Rahman Wijaya, Wawa Wikusna","doi":"10.1109/ICoDSA55874.2022.9862888","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862888","url":null,"abstract":"Meat is a source of food that contains many nutrients. The nutritional content of meat consists of fat, calories, trans fat, saturated fat, calcium, protein, vitamin D, vitamin B6, vitamin B12, and magnesium. Due to its good nutritional content, the demand for meat in Indonesia has increased. However, there are problems with meat health. Meat is prone to spoilage and is quickly contaminated with microbes. The microbial population can spoil or spoil the meat. Checking the feasibility of meat is usually done by looking at the texture of the meat traditionally. However, this method is less effective in assessing the feasibility of meat. Therefore, another method is used to determine the feasibility of meat, namely using the Electronic Nose (e-nose) with the Neural Network (NN) algorithm. Because by using an e-nose, that can find out the smell or smell of decent meat. They are applying the NN algorithm for classification to work in a structured manner on each component needed to determine meat quality. These results can help people to get the meat of good quality. The experiment was carried out using a dataset that had a total of 2220 data. The experimental results show that using the NN algorithm with the e-nose sensor gets an accuracy of 0.92.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115247101","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862881
Eprian Junan Setianto, Esmeralda Contessa Djamal, Fikri Nugraha, Fatan Kasyidi
E-commerce products need to be measured by reader responses as a more objective evaluation. Some of them are through emotion expression identification or eye-tracking. Using these two variables from video capture provides a more thorough evaluation of the response to interest and emotion. This study proposes a spatial-temporal multi-networks method using Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) from video for 60 seconds. The results showed that two classes of emotional expression and four directions of eye-tracking gave better accuracy, namely 95.83%, compared to three classes of emotion and four directions of eye-tracking, which was 91.67%. Experiments also show that using CNN-LSTM significantly increased accuracy, while the weight correction technique does not have much effect. The evaluated F1 score shows the consistency of the proposed model.
{"title":"Eye Tracking and Emotion Recognition Using Multiple Spatial-Temporal Networks","authors":"Eprian Junan Setianto, Esmeralda Contessa Djamal, Fikri Nugraha, Fatan Kasyidi","doi":"10.1109/ICoDSA55874.2022.9862881","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862881","url":null,"abstract":"E-commerce products need to be measured by reader responses as a more objective evaluation. Some of them are through emotion expression identification or eye-tracking. Using these two variables from video capture provides a more thorough evaluation of the response to interest and emotion. This study proposes a spatial-temporal multi-networks method using Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) from video for 60 seconds. The results showed that two classes of emotional expression and four directions of eye-tracking gave better accuracy, namely 95.83%, compared to three classes of emotion and four directions of eye-tracking, which was 91.67%. Experiments also show that using CNN-LSTM significantly increased accuracy, while the weight correction technique does not have much effect. The evaluated F1 score shows the consistency of the proposed model.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"363 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122843865","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862833
Mahendra Data, M. Aritsugi
Most machine learning models used in network intrusion detection system (IDS) studies are batch models which require all targeted intrusions to be present in the training data. This approach is slow because computer networks produce massive amounts of data. Furthermore, new network intrusion variants continuously emerge. Retraining the model using these extensive and evolving data takes time and resources. This study proposes AB-HT: an ensemble incremental learning algorithm for IDSs. AB-HT utilizes incremental Adaptive Boosting (AdaBoost) and Hoeffding Tree algorithms. AB-HT model could detect new intrusions without retraining the model using old training data. Thus, it could reduce the computational resources needed to retrain the model while maintaining the model’s performance. We compared it to an AdaBoost-Decision Tree model, a batch learning model, to analyze the effectiveness of the incremental learning approach. Then we compared it to other incremental learning models, the Hoeffding Tree (HT) and Hoeffding Anytime Tree (HATT) models. The experimental results showed that the proposed incremental model had shorter training times than the AdaBoost-Decision Tree model in the long run. Also, on average, the AB-HT model has 18% higher F1-score values than the HT and HATT models. These advantages show that the AB-HT algorithm has promising potential to be used in the IDS field.
{"title":"AB-HT: An Ensemble Incremental Learning Algorithm for Network Intrusion Detection Systems","authors":"Mahendra Data, M. Aritsugi","doi":"10.1109/ICoDSA55874.2022.9862833","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862833","url":null,"abstract":"Most machine learning models used in network intrusion detection system (IDS) studies are batch models which require all targeted intrusions to be present in the training data. This approach is slow because computer networks produce massive amounts of data. Furthermore, new network intrusion variants continuously emerge. Retraining the model using these extensive and evolving data takes time and resources. This study proposes AB-HT: an ensemble incremental learning algorithm for IDSs. AB-HT utilizes incremental Adaptive Boosting (AdaBoost) and Hoeffding Tree algorithms. AB-HT model could detect new intrusions without retraining the model using old training data. Thus, it could reduce the computational resources needed to retrain the model while maintaining the model’s performance. We compared it to an AdaBoost-Decision Tree model, a batch learning model, to analyze the effectiveness of the incremental learning approach. Then we compared it to other incremental learning models, the Hoeffding Tree (HT) and Hoeffding Anytime Tree (HATT) models. The experimental results showed that the proposed incremental model had shorter training times than the AdaBoost-Decision Tree model in the long run. Also, on average, the AB-HT model has 18% higher F1-score values than the HT and HATT models. These advantages show that the AB-HT algorithm has promising potential to be used in the IDS field.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129704541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862826
Saskia Putri Ananda, Z. Baizal
Digital health solutions can be done in various ways, one of which is by searching for information on the internet. However, when someone searches on a search engine, the videos that are displayed are only videos based on keywords, without considering what kind of videos the user likes. Meanwhile, when searching for videos on YouTube, the recommended videos are only videos found on YouTube, so the range of recommended videos is limited. To overcome this problem, we build a web-based video recommender system about diseases that is more organized with a wider range of videos taken from YouTube and Vimeo. In addition, the system not only recommends videos based on the searched keywords but also recommends videos based on videos that are liked by users. The YouTube and Vimeo APIs are used to retrieve videos about the disease being searched for. We use content-based filtering for the recommendation process. Keyword-based vector space does some tasks: 1) converts the title and description of a video into a vector space, 2) calculates the cross product of the term frequency, 3) determines the proximity of the title using cosine similarity. The test results show that the average performance is 92.67% according to the purpose of the recommendation system made, namely novelty and relevance.
{"title":"Diseases Video Recommender System using Keyword-Based Vector Space on Youtube and Vimeo","authors":"Saskia Putri Ananda, Z. Baizal","doi":"10.1109/ICoDSA55874.2022.9862826","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862826","url":null,"abstract":"Digital health solutions can be done in various ways, one of which is by searching for information on the internet. However, when someone searches on a search engine, the videos that are displayed are only videos based on keywords, without considering what kind of videos the user likes. Meanwhile, when searching for videos on YouTube, the recommended videos are only videos found on YouTube, so the range of recommended videos is limited. To overcome this problem, we build a web-based video recommender system about diseases that is more organized with a wider range of videos taken from YouTube and Vimeo. In addition, the system not only recommends videos based on the searched keywords but also recommends videos based on videos that are liked by users. The YouTube and Vimeo APIs are used to retrieve videos about the disease being searched for. We use content-based filtering for the recommendation process. Keyword-based vector space does some tasks: 1) converts the title and description of a video into a vector space, 2) calculates the cross product of the term frequency, 3) determines the proximity of the title using cosine similarity. The test results show that the average performance is 92.67% according to the purpose of the recommendation system made, namely novelty and relevance.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127943650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862894
Muhammad Khanif Khafidli, A. Choiruddin
The COVID-19 pandemic has impacted many sectors. For example, in the aviation sector, flight traffic went down drastically with no certainty of being recovered. This calls for a methodology to predict the flight traffic to provide strategic planning on flight schedules operational, route structuring, and flight navigation service cost determination. However, current developments mainly focus on flight traffic forecasting based on historical data without considering external factors. In this study, we propose the Long Short-Term Memory (LSTM) technique to forecast flight traffic in Indonesia involving external variables such as macroeconomic variables and Google Trends. LSTM is proposed because of its flexibility to model non-linear time series data and has a good reputation for predictive accuracy. We first select a few among Google Trends and macroeconomic variables using nonlinearity analysis and cross-correlation function (CCF). We then employ the selected variables to forecast the flight traffic and compare it to the one using only historical flight traffic data. Our results concluded, based on the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE), that the model involving google trend outperforms the other three models, i.e., the model with only historical data, the model with macroeconomics, and the model with both macroeconomic and Google Trends. It is because, in this digital era, Google Trends can reflect population psychology in an up-to-date manner.
{"title":"Forecast of Aviation Traffic in Indonesia Based on Google Trend and Macroeconomic Data using Long Short-Term Memory","authors":"Muhammad Khanif Khafidli, A. Choiruddin","doi":"10.1109/ICoDSA55874.2022.9862894","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862894","url":null,"abstract":"The COVID-19 pandemic has impacted many sectors. For example, in the aviation sector, flight traffic went down drastically with no certainty of being recovered. This calls for a methodology to predict the flight traffic to provide strategic planning on flight schedules operational, route structuring, and flight navigation service cost determination. However, current developments mainly focus on flight traffic forecasting based on historical data without considering external factors. In this study, we propose the Long Short-Term Memory (LSTM) technique to forecast flight traffic in Indonesia involving external variables such as macroeconomic variables and Google Trends. LSTM is proposed because of its flexibility to model non-linear time series data and has a good reputation for predictive accuracy. We first select a few among Google Trends and macroeconomic variables using nonlinearity analysis and cross-correlation function (CCF). We then employ the selected variables to forecast the flight traffic and compare it to the one using only historical flight traffic data. Our results concluded, based on the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE), that the model involving google trend outperforms the other three models, i.e., the model with only historical data, the model with macroeconomics, and the model with both macroeconomic and Google Trends. It is because, in this digital era, Google Trends can reflect population psychology in an up-to-date manner.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"172 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120941306","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862919
Ridhwan Nashir, E. B. Setiawan, D. Adytia
Sentiment towards a company is suspected of influencing the company's stock price movement. The sentiment is gathered from Twitter, Youtube, Facebook with some news media such as Consumer News and Business Channel (CNBC), Kontan, Detik, Cable News Network (CNN), Stockbit, and Liputan6 which discussed Bank Mandiri. Word2Vec is used to reduce vocabulary errors in sentiment analysis using word embedding. The Word2Vec model was built using the combined corpus of Wikipedia articles and scraped data with a total of 474,277 lines of text data. This study indicates that the correlation between sentiment and stock movements of Bank Mandiri has a positive correlation with a low relationship, indicated by the Spearman Rank test coefficient value of 0.138 and 0.123 for positive and negative sentiment, respectively. The Naïve Bayes-Support Vector Machine (NBSVM) classification model outperforms the Naïve Bayes and Support Vector Machine methods, where the baseline NBSVM gets an accuracy of 64.67%, and after the feature expansion process, the accuracy becomes 70.42%, an increase of 5.75%. This study proves there is a correlation between sentiment and the movement of Bank Mandiri's shares, and Word2Vec feature expansion can increase the model's accuracy.
{"title":"The Influence of Sentiment on the Movement of Bank Mandiri (BMRI) Stock Price with Word2Vec Feature Expansion and the Naïve Bayes-Support Vector Machine (NBSVM) Classifier","authors":"Ridhwan Nashir, E. B. Setiawan, D. Adytia","doi":"10.1109/ICoDSA55874.2022.9862919","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862919","url":null,"abstract":"Sentiment towards a company is suspected of influencing the company's stock price movement. The sentiment is gathered from Twitter, Youtube, Facebook with some news media such as Consumer News and Business Channel (CNBC), Kontan, Detik, Cable News Network (CNN), Stockbit, and Liputan6 which discussed Bank Mandiri. Word2Vec is used to reduce vocabulary errors in sentiment analysis using word embedding. The Word2Vec model was built using the combined corpus of Wikipedia articles and scraped data with a total of 474,277 lines of text data. This study indicates that the correlation between sentiment and stock movements of Bank Mandiri has a positive correlation with a low relationship, indicated by the Spearman Rank test coefficient value of 0.138 and 0.123 for positive and negative sentiment, respectively. The Naïve Bayes-Support Vector Machine (NBSVM) classification model outperforms the Naïve Bayes and Support Vector Machine methods, where the baseline NBSVM gets an accuracy of 64.67%, and after the feature expansion process, the accuracy becomes 70.42%, an increase of 5.75%. This study proves there is a correlation between sentiment and the movement of Bank Mandiri's shares, and Word2Vec feature expansion can increase the model's accuracy.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133264183","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-07-06DOI: 10.1109/ICoDSA55874.2022.9862938
R. V. Priyanka Hafidz, Amelia Setiawan
With digital developments that continue to occur, it can affect several parts of the institution or company by changing the system from manual to online. With this progress, government institutions have also begun to implement a computerized system and payroll is one of the sections that is affected by it. For example, payroll which is usually given in person, can now be sent via bank transfer. This study was conducted to analyze the quality of information, system quality, and information system security on user satisfaction in the payroll section of a government institution, namely, the Center for Financial Transaction Reports and Analysis. This study will use data that has been processed in two ways. The first is one of the functions of Microsoft Excel, namely data analysis and the second uses SEM PLS analysis to test the three pre-determined hypotheses. The results of hypothesis testing indicate that the quality of information and information system security affect user satisfaction significantly, while system quality does not substantially affect user satisfaction. The limitation of this research is the limited number of employees in the payroll section of the Financial Transaction Reports and Analysis Center. Suggestions for further research are to use a more general section that has more employees.
{"title":"What Affects User Satisfaction of Payroll Information Systems?","authors":"R. V. Priyanka Hafidz, Amelia Setiawan","doi":"10.1109/ICoDSA55874.2022.9862938","DOIUrl":"https://doi.org/10.1109/ICoDSA55874.2022.9862938","url":null,"abstract":"With digital developments that continue to occur, it can affect several parts of the institution or company by changing the system from manual to online. With this progress, government institutions have also begun to implement a computerized system and payroll is one of the sections that is affected by it. For example, payroll which is usually given in person, can now be sent via bank transfer. This study was conducted to analyze the quality of information, system quality, and information system security on user satisfaction in the payroll section of a government institution, namely, the Center for Financial Transaction Reports and Analysis. This study will use data that has been processed in two ways. The first is one of the functions of Microsoft Excel, namely data analysis and the second uses SEM PLS analysis to test the three pre-determined hypotheses. The results of hypothesis testing indicate that the quality of information and information system security affect user satisfaction significantly, while system quality does not substantially affect user satisfaction. The limitation of this research is the limited number of employees in the payroll section of the Financial Transaction Reports and Analysis Center. Suggestions for further research are to use a more general section that has more employees.","PeriodicalId":339135,"journal":{"name":"2022 International Conference on Data Science and Its Applications (ICoDSA)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115539285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：