Claim Missing Document
Check
Articles

Using Social Media Data to Monitor Natural Disaster: A Multi Dimension Convolutional Neural Network Approach with Word Embedding Mohammad Reza Faisal; Irwan Budiman; Friska Abadi; Muhammad Haekal; Mera Kartika Delimayanti; Dodon Turianto Nugrahadi
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 6 No 6 (2022): Desember 2022
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v6i6.4525

Abstract

Social media has a significant role in natural disaster management, namely as an early warning and monitoring when natural disasters occur. Artificial intelligence can maximize the use of natural disaster social media messages for natural disaster management. The artificial intelligence system will classify social media message texts into three categories: eyewitness, non-eyewitness and don't-know. Messages with the eyewitness category are essential because they can provide the time and location of natural disasters. A common problem in text classification research is that feature extraction techniques ignore word meanings, omit word order information and produce high-dimensional data. In this study, a feature extraction technique can maintain word order information and meaning by using three-word embedding techniques, namely word2vec, fastText, and Glove. The result is data with 1D, 2D, and 3D dimensions. This study also proposes a data formation technique with new features by combining data from all word embedding techniques. The classification model is made using three Convolutional Neural Network (CNN) techniques, namely 1D CNN, 2D CNN and 3D CNN. The best accuracy results in this study were in the case of earthquakes 78.33%, forest fires 81.97%, and floods 78.33%. The calculation of the average accuracy shows that the 2D and 3D v1 data formation techniques work better than other techniques. Other results show that the proposed technique produces better average accuracy.
Efek Transformasi Wavelet Diskrit Pada Klasifikasi Aritmia Dari Data Elektrokardiogram Menggunakan Machine Learning Dodon Turianto Nugrahadi; Tri Mulyani; Dwi Kartini; Rudy Herteno; Mohammad Reza Faisal; Irwan Budiman; Friska Abadi
JURNAL MEDIA INFORMATIKA BUDIDARMA Vol 7, No 1 (2023): Januari 2023
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/mib.v7i1.4859

Abstract

Arrhythmia is one of the abnormalities of the heart rhythm, and some patients who suffer from arrhythmia do not feel any symptoms. Automating the early detection of arrhythmia is necessary by using an electrocardiogram. Previous research that had been done conducted classifications using several methods of data mining. In this research, the transformation for processing signals used is Discrete Wavelet Transformation, where a filtering process occurs that separates signals into high and low-frequency signals without losing the information from signals and is carried out with a two-level decomposition. After that, data normalization was performed using min-max normalization and was put into the model classification using the Support Vector Machine method with a Gaussian Radial Basis Function kernel of Naïve Bayes and K-Nearest Neighbor. Each data that was being used consisted of 140 data with a total of 35 data for each label. This research shows that at level 1 decomposition, the highest accuracy was obtained at db7 for the classification using Support Vector Machine with an accuracy of 73,57%, 68,57% for Naïve Bayes, K-Nearest Neighbor with k=3 resulting in an accuracy of 59,64%, and K-Nearest Neighbor with k=5 resulting in an accuracy of 63,57% while at level 2 decomposition the highest accuracy was obtained at db6 dan db8 for the classification using Support Vector Machine with an accuracy of 70,71%, 67,50% for Naïve Bayes, K-Nearest Neighbor with k=3 resulting in an accuracy of 66,07%, and K-Nearest Neighbor with k=5 resulting in an accuracy of 65%. From this research, it can be concluded that the highest accuracy is produced by decomposition level 1 using Support Vector Machine classification and that the Daubechies wavelet type has better results than the Haar wavelet.
PENGARUH SOFTWARE METRIK PADA KINERJA KLASIFIKASI CACAT SOFTWARE DENGAN ANN Achmad Zainudin Nur; Mohammad Reza Faisal; Friska Abadi; Irwan Budiman; Rudy Herteno
Journal of Data Science and Software Engineering Vol 1 No 01 (2020)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (180.324 KB) | DOI: 10.20527/jdsse.v1i01.5

Abstract

Software Defect Prediction has an important role in quality software. This study uses 12 D datasets from NASA MDP which then features a selection of metrics categories software. Feature selection is performed to find out metrics software which are influential in predicting defects software. After the feature selection of the metric software category, classification will be performed using the algorithm Artificial Neural Network and validated with 5-Fold Cross Validation. Then conducted an evaluation with Area Under Curve (AUC), From datasets D” 12 NASA MDP that were evaluated with AUC, PC4, PC1 and PC3 datasets obtained the best AUC performance values. Each value is 0.915, 0.828, and 0.826 using the algorithm Artificial Neural Network.
Penyeleksian Calon Karyawan Menggunakan Metode Pembobotan Shannon Entropy dan Metode ARAS Halimah; Dwi Kartini; Friska Abadi; Irwan Budiman; Muliadi
Journal of Data Science and Software Engineering Vol 1 No 01 (2020)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (220.092 KB) | DOI: 10.20527/jdsse.v1i01.7

Abstract

This study discusses the selection of prospective employees using the Shannon Entropy weighting method and the Additive Ratio Assessment (ARAS) method which aims to determine the accuracy of the results obtained from the method. The Shannon Entropy method is a weighting method that assigns criteria weights based on the calculation of alternative employee selection data and the Additive Ratio Assessment (ARAS) method is a ranking method that has a utility function. Testing the data in this study using the Mean Absolute Error (MAE) method to get system accuracy results. Based on testing conducted using 6 criteria and 56 alternative data for prospective employees, the accuracy of the method used was 85.34%.
IDENTIFIKASI PESAN SAKSI MATA PADA BENCANA KEBAKARAN HUTAN MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK Rinaldi; Mohammad Reza Faisal; Muhammad Itqan Mazdadi; Radityo Adi Nugroho; Friska Abadi
Journal of Data Science and Software Engineering Vol 2 No 02 (2021)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (370.011 KB)

Abstract

Social media, one of which is Twitter, is a medium for disseminating information that is growing rapidly at this time. The advantage of Twitter which has such a huge impact is its speed in spreading news and information that is happening. One of the information that is often reported through social media is information about natural disasters. Therefore, a lot of research on sensor social networks has been carried out by researchers using data from social media with the aim of obtaining valid data for the disaster emergency response process. In this study, the classification of eye witness messages for forest fires was carried out using Convolutional Neural Network and feature extraction Word2Vec with dimensions of 100. Twitter data used amounted to 3000 data and divided into 3 classes, namely eyewitnesses, non-eyewitnesses, and unknowns. The research was conducted to determine the accuracy performance obtained from testing using several types of configurations hyperparameter. Based on the results of the tests carried out, the best accuracy value was 81.97%.
OPTIMASI NILAI N PADA SINGLE MOVING AVERAGE (SMA) DENGAN PARTICLE SWARM OPTIMIZATION (PSO) STUDI KASUS SAHAM BRI Rahman Hadi Rahman; Irwan Budiman; Friska Abadi; Andi Farmandi; Muliadi
Journal of Data Science and Software Engineering Vol 2 No 03 (2021)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (456.928 KB)

Abstract

The stock market is a promising business area. The potential to obtain high returns in a fairly short time is one of the main attractions of this business. Prediction of stock prices has become a very interesting and challenging thing for researchers and academics, recently it was found that stock prices can be predicted with a certain degree of accuracy. Single Moving Average (SMA) is one method for predicting time series data. However, the N value in SMA needs to be optimized in order to get the N value with optimal results at the SMA and get accurate results. The Particle Swarm Optimization Algorithm is implemented to find out the best N value in the Single Moving Average methodwhich is more optimal. SMA+PSO and SMA are calculated using the initial N values ​​of 3,5,7,9,11. So the results of this study are SMA with an accuracy of 97.98464% and for SMA+PSO with an accuracy of 98.15442% . The test results from this study are the influence of PSO on SMA in increasing accuracy in determining the best N value.
SISTEM PENDUKUNG KEPUTUSAN PEMILIHAN POHON UNTUK RESTORASI LAHAN BEKAS KEBAKARAN DENGAN METODE ANALYTIC HIERARCHY PROCESS (AHP) DAN SIMPLE MULTI ATTRIBUTE RATING TECHNIQUE EXPLOITING RANKS (SMARTER) Muhammad Denny Ersyadi Rahman; Muliadi; Rudy Herteno; Dwi Kartini; Friska Abadi
Journal of Data Science and Software Engineering Vol 2 No 03 (2021)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (449.162 KB)

Abstract

Utilization or use of forest and land areas that are not in accordance with conservation principles can cause critical land to occur. Critical land is land inside or outside the forest area that has been damaged, so that it can cause loss or decrease in its function. The lack of knowledge of some people on critical land and the selection of inappropriate plant types sometimes makes the condition of burnt land increasingly become one of the obstacles for the Forest and Land Rehabilitation Program (RHL). Statistical data analysis can be used in the data processing process to become valuable information for the system. Applying statistical analysis methods in making decisions in selecting statistical data that has several criteria. This research is focused on the application of the Analytical Hierarchy Process (AHP) method to see a comparison of criteria. The SMARTER (Simple Multi Attribute Rating Technique Exploiting Rank) method is very suitable to be used to overcome the many alternatives that will be given to different soil samples later. In short, each final weight that affects the alternative is calculated with the results of the alternative assessment, so that the utility value of each alternative is obtained. From the research of the Analytical Hierarchy Process (AHP) and Simple Multi Attribute Rating Technique Exploiting Rank (SMARTER) method, the results of the Balangeran vegetation are obtained as the main recommendation with the greatest utility value, namely 1.321668.
SOLUSI KLASIFIKASI DATA TIDAK SEIMBANG DENGAN PENDEKATAN BERBASIS COMBINATION OF OVERSAMPLING AND UNDERSAMPLING Riza Susanto Banner; Irwan Budiman; Dodon Turianto Nugrahadi; M. Reza Faisal; Friska Abadi
Journal of Data Science and Software Engineering Vol 3 No 01 (2022)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (345.141 KB)

Abstract

This study applies the Combination of Oversampling and Undersampling method to deal with class imbalances. Researchers do Preprocessing to normalize the attributes used for prediction, then divide the training data and testing data. Researchers resampled unbalanced data using Oversampling, Undersampling and a combination of Oversampling and Undersampling. The results of the classification with the experimental data class balancing approach, the best classification performance is the combination of Oversampling and Undersampling classified by the k-Nearest Neighbor (KNN) method with an accuracy of 0.8672; sensitivity of 0.9000; specificity of 0.3750; and AUC of 0.6651042. Classification with Oversampling has performance results, namely accuracy of 0.875; sensitivity of 0.9250; specificity of 0.1250; and AUC of 0.6078125, while Undersampling classification has classification performance, namely accuracy of 0.3438; sensitivity of 0.33333; specificity of 0.50000; and AUC of 0.3645833.
COMPARATIVE ANALYSIS OF FUZZY TIME SERIES METHOD WITH FUZZY TIME SERIES MARKOV CHAIN ON RAINFALL FORECAST IN SOUTH KALIMANTAN M Kevin Warendra; Irwan Budiman; Rudy Herteno; Dodon Turianto Nugrahadi; Friska Abadi
Journal of Data Science and Software Engineering Vol 3 No 01 (2022)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (1219.79 KB)

Abstract

Abstract Time series data (TS) is a type of data that is collected according to the order of time within a certain time span. Time Series data analysis is one of the statistical procedures applied to predict the probability structure of future conditions for decision making. FTS (FTS) is a data forecasting method that uses fuzzy principles as its basis. Forecasting systems with FTS capture patterns from past data and then use them to project future data. FTS Markov Chain is a new concept that was first proposed by Tsaur, in his research to analyze the accuracy of the prediction of the Taiwan currency exchange rate with the US dollar. In his research, Tsaur combines the FTS method with Markov Chain, The merger aims to obtain the greatest probability using a transition probability matrix. The results obtained from this research are tests with the best number of presentation values ​​from FTS Markov Chain with FTS, resulting in different accuracy values ​​depending on the two methods. The best accuracy performance is obtained by the Markov Chain FTS method with an error value of 1.6% and an accuracy value of 98.4% and for FTS with an error value of 7.4% and an accuracy value of 92.6%. produce different accuracy values ​​depending on the two methods. The best accuracy performance is obtained by the Markov Chain FTS method with an error value of 1.6% and an accuracy value of 98.4% and for FTS with an error value of 7.4% and an accuracy value of 92.6%. produce different accuracy values ​​depending on the two methods. The best accuracy performance is obtained by the Markov Chain FTS method with an error value of 1.6% and an accuracy value of 98.4% and for FTS with an error value of 7.4% and an accuracy value of 92.6%.
SISTEM PEMANTAUAN LOKASI PEGAWAI ULM BERBASIS PRESENSI BERGERAK Ahmad Juhdi; Radityo Adi Nugroho; Friska Abadi; Andi Farmadi; Rudy Herteno
Journal of Data Science and Software Engineering Vol 3 No 02 (2022)
Publisher : Fakultas MIPA Universitas Lambung Mangkurat

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (558.312 KB)

Abstract

ULM attendance is usually done in each faculty using a fingerprint-based attendance machine. However, fingerprint-based presence during the pandemic is very dangerous due to the COVID-19 outbreak which allows the spread of the virus to be transmitted through finger intermediaries who use the presence machine simultaneously. As well as the existence of a letter prohibiting going home issued by the MENPENRB regarding "Restrictions on traveling activities outside the region or homecoming activities or leave for ASN in an effort to prevent the spread of Covid-18". In this study, we use a smartphone-based electronic system to overcome fingerprint-based attendance problems so that we can get an increase in terms of costs, and minimize the spread of the COVID-19 outbreak. By knowing the level of profit achieved through investment in the application development that the researcher has proposed, it is necessary to conduct a feasibility study (Feasibility Analysis) as a tool in drawing conclusions about what will be done electronically, a comparison will be made against the implementation of attendance in the previous year. The operational costs required are Rp. 27,665,070, while the costs incurred for application development are Rp. 1,613,666, it can be seen that there is an implementation cost savings of Rp. 26,051,404, when operational cost savings are included in the economic feasibility study, the Return on Investment (ROI) and Break-Event Point (BEP) values since the first year the application was implemented showed a positive value. Until the fourth year, ROI and BEP entered the feasible criteria so that from an Economic Feasibility perspective it can be seen that the application is economically feasible. And the application that is made is able to provide convenience in using the application as evidenced by validity and reliability tests.