Claim Missing Document
Check
Articles

Found 38 Documents
Search

Clustering Gempabumi di Wilayah Regional VII Menggunakan Pendekatan DBSCAN Arafat, Ihsan Bagus Fahad; Hariyadi, Mokhamad Amin; Santoso, Irwan Budi; Crysdian, Cahyo
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 10 No 4: Agustus 2023
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25126/jtiik.2024106918

Abstract

Wilayah Regional VII meliputi Jawa Tengah, Yogyakarta, dan Jawa Timur merupakan wilayah tektonik yang aktif karena terletak di wilayah zona subduksi lempeng Indo-Australia dan Eurasia serta terdapat beberapa patahan aktif di daratan. Oleh karena itu, perlu dilakukan klasifikasi gempabumi untuk memetakan zona rawan gempabumi berdasarkan sumbernya di wilayah Regional VII berdasarkan kesamaan atribut salah satunya adalah berdasarkan karakteristik gempabumi dari sumber yang sama. Pada penelitian ini digunakan pendekatan algoritma Unsupervised Learning Clustering berbasis kepadatan yaitu, Density Based Spatial Clustering of Application with Noise atau DBSCAN, algoritma ini membutuhkan parameter input epsilon (ε) dan MinPts. Data yang digunakan pada penelitian ini adalah data gempabumi wilayah Regional VII tahun 2017 hingga 2021 yang diperoleh dari BMKG. Selanjutnya, proses clustering dilakukan dengan membagi data gempabumi berdasarkan periode yaitu periode tahunan dan periode lima tahun dengan tujuan untuk mengetahui pola cluster berdasarkan periode waktu. Hasil yang terbentuk selanjutnya dievaluasi menggunakan Silhouette Coefficient serta dibandingkan dengan peta Seismisitas Jawa yang telah ada dari katalog PuSGeN 2017. Hasil clustering menggunakan DBSCAN diperoleh jumlah cluster sebanyak 2 hingga 6 cluster dengan nilai Silhouette Coefficient terendah sebesar 0.270 untuk periode T5_2017-2021 dan tertinggi sebesar 0.499 untuk periode T1_2020. AbstractRegional VII area covering Central Java, Yogyakarta and East Java is an active tectonic region because it is located in the subduction zone of the Indo-Australian and Eurasian plates and there are several active faults on land. Therefore, it is necessary to classify earthquakes to map earthquake-prone zones based on their sources in Regional VII area based on the similarity of attibutes, based on the characteristics of earthquakes from the same source. In this study, a density-based Unsupervised Learning Clustering algorithm approach was used namely, Density Based Spatial Clustering of Application with Noise or DBSCAN, this algorithm requires the input parameters epsilon (ε) and MinPts. The data used in this study are earthquake data for Regional VII from 2017 to 2021 obtained from the BMKG. Then, the clustering process is carried out by dividing earthquake data based on the period, namely the annual period and the five-year period with the aim of knowing the pattern of cluster based on the time period. The results are then evaluated using the Sillhouette Coefficient and compared with the existing Java Seismicity map from the 2017 PuSGeN catalog. Clustering results using DBSCAN obtained a number of clusters of 2 to 6 clusters with the lowest Silhouette Coefficient value is 0.270 for the T5_2017-2021 period and the highest is 0.499 for the T1_2020 period.  
Chat Bot News Analysis to Support Education Services Using Neural Network Khamidatullailiyah, Yayang; Crysdian, Cahyo; Fatchurrohman, Fatchurrohman
International Journal of Information Systems and Technology Vol. 1 No. 02 (2025): International Journal of Information System and Technology (IJOINT)
Publisher : oneamrd (One a Mllions Dreams Corporation)

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

During the Covid-19 pandemic, all learning is held online, so all lecture activities up to student administration must be done online. To make it easier for students to obtain information about academics and administration, a chatbot feature is needed which can provide information while communicating two-way to students as users. Chatbots can provide services practically, quickly, and responsively. In order for the chatbot to provide answers that match user expectations, the question sentences that enter the system must be classified properly and correctly. This study applies the Neural Network method to classify answers on chatbots. Neural Networks are used in research methods because they can build models easily and can be used to classify text with a high level of accuracy. To measure the performance of the chatbot system in providing appropriate answers, an evaluation is carried out by calculating the accuracy, precision, recall, and f-measuring values ​​using a confusion matrix. The results of the study show that the Neural Network method built on the chatbot system in classifying answers can run well with an accuracy value of 99.21%, precision of 88.09%, recall of 88.09%, and f-measure of 88.09%.
Perbandingan Feature extraction TF-IDF dan BOW Untuk Analisis Sentimen Berbasis SVM Putra, Kurniawan Tri; Hariyadi, Mokhamad Amin; Crysdian, Cahyo
Jurnal Cahaya Mandalika ISSN 2721-4796 (online) Vol. 3 No. 2 (2022)
Publisher : Institut Penelitian Dan Pengambangan Mandalika Indonesia (IP2MI)

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Dengan adanya transformasi society 5.0 pegaruh paling besar yang bisa dirasakan saat ini adalah berkembang pesatnya jumlah data yang ada di seluruh dunia baik yang bermanfaat secara langsung maupun data yang tidak bermanfaat secara langsung atau dikenal dengan istilah big data, dengan adanya sumber big data tersebut banyak peneliti-peneliti yang memanfaatkanya menjadi suatu data yang berharga dan berguna jika diproses dan diolah dengan cara yang baik dan benar salah satunya adalah dengan tujuan analisis sentimen. Pada permasalahan yang ada penelitian ini bertujuan untuk mencari dan mendapatkan alur dan teknik yang benar serta memiliki hasil optimal pada pengolahan data teks dengan tujuan analisis sentimen dengan membandingakan penerapan TF-IDF dan BOW yang menggunakan metode SVM. Pada penelitian analisis sentimen menggunakan data teks bersumber dari aplikasi media social twitter hasil yang didapatkan adalah pada penerapan teknik TF-IDF yang dipadukan dengan metode SVM memiliki hasil yang lebih baik dengan nilai Accuracy 86%, Precission 85%, Recall 85% dan F1-Score 85% sedangkan penerapan teknik BOW yang dipadukan metode SVM hanya unggul pada nilai Recall sebesar 89%.
PREDIKSI TINGKAT KEPERCAYAAN MASYARAKAT TERHADAP PILPRES 2024 MENGGUNAKAN TF-IDF DAN BOW MENGGUNAKAN METODE SVM Mustaqim, Eka Rifut Nur; Pagalay, Usman; Crysdian, Cahyo
Jurnal Cahaya Mandalika ISSN 2721-4796 (online) Vol. 5 No. 1 (2024)
Publisher : Institut Penelitian Dan Pengambangan Mandalika Indonesia (IP2MI)

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Dalam era modern ini, dunia maya telah menjadi salah satu aspek yang tak terpisahkan dari kehidupan sehari-hari kita. Dunia maya, atau internet, adalah hasil dari kemajuan teknologi informasi yang telah merevolusi dunia selama beberapa dekade terakhir. Namun, lebih dari sekadar teknologi, ini telah menjadi sebuah ekosistem yang hidup, dihuni oleh miliaran orang yang terhubung, menciptakan dan mengonsumsi informasiPrediksi pada pemanfaatan big data ini dengan cara kerja mencari dan mengolah data dari segala bentuk ekspresi atau keadaan yang sedang atau telah dialami seseorang user yang diluangkan dalam bentuk teks kedalam media sosial, Prediksi tidak harus memberikan jawaban secara pasti kejadian yang akan terjadi, melainkan berusaha untuk mencari jawaban sedekat mungkin yang akan terjadi.Berdasarkan pada permasalahan yang telah dibahas beberapa teknik yang paling umum dan sering digunakan dalam feature extraction TF-IDF dan BOW, dikarenakan kedua teknik tersebut sangat bersaing serta berperan baik dan sama-sama digunakan untuk merepresentasikan numerik dari data teks serta memiliki kekurangan dan kelebihan masing masing. Pada penelitian kali ini akan membandingkan kedua metode tersebut yan dipadukan dengan menggunakan metode SVM, untuk model penelitian TF-IDF dengan menggunakan metode SVM mendapatkan hasil Accurasi sebesar 85%, hasil nilai precission sebesar 84%, hasil Recall sebesar 83% dan untuk hasil F1-Score sebesar 83%, sedangkan penelitian menggunakan teknik BOW dengan metode SVM mendapatkan hasil Accurasi sebesar 84%, hasil nilai precission sebesar 79%, hasil Recall sebesar 89% dan untuk hasil F1-Score sebesar 83%.
Rainfall Prediction Using Attention-Based LSTM Architecture Romadhani, Ahmad; Santoso, Irwan Budi; Crysdian, Cahyo
JURNAL RISET KOMPUTER (JURIKOM) Vol. 12 No. 3 (2025): Juni 2025
Publisher : Universitas Budi Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30865/jurikom.v12i3.8727

Abstract

This study addresses the challenge of accurately predicting rainfall in regions with complex climate dynamics, such as Malang Regency, East Java. It evaluates the performance of a Long Short-Term Memory (LSTM) model enhanced with the Bahdanau Attention Mechanism, comparing it with a Standard LSTM model in forecasting daily rainfall based on historical weather parameters including average temperature, relative humidity, sunshine duration, and wind speed. Using daily data from BMKG covering 2000 to 2023, both models underwent a structured machine learning process including data preprocessing, feature selection, model training, and evaluation. The Attention-Based LSTM consistently outperformed the Standard LSTM, particularly in handling rainfall anomalies, achieving an MSE of 0.00800 and RMSE of 0.08948, compared to 0.00817 and 0.09039 respectively for the Standard LSTM. These results demonstrate that integrating Bahdanau Attention improves the model’s focus on relevant temporal features, enhancing prediction accuracy and robustness. The architecture consisting of two LSTM cells combined with the attention mechanism effectively captures complex sequential patterns that the standard model tends to overlook. This research highlights the potential of attention mechanisms in time series weather prediction, contributing to more reliable early warning systems, adaptive agricultural strategies, and disaster risk reduction frameworks. Future work could explore hybrid models or incorporate additional weather features to further improve performance and generalization.
E-commerce Product Review Classification using Neural Network-Based Approach Ihtada, Fahrendra Khoirul; Abidin, Zainal; Crysdian, Cahyo
JOIV : International Journal on Informatics Visualization Vol 9, No 6 (2025)
Publisher : Society of Visual Informatics

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62527/joiv.9.6.2845

Abstract

E-commerce has become an integral part of how people shop, with the rise of customer reviews on various platforms. These reviews provide important insights into product, customer service, and delivery. The growing volume of e-commerce reviews makes manual sorting time-consuming and error-prone for business owners. This study aims to classify e-commerce reviews into three categories: product, customer service, and delivery. The data was collected from e-commerce customer reviews on Tokopedia and labeled using crowdsourcing for ground truth. To classify the reviews, a Neural Network is performed with various numbers of node and learning rate. TF-IDF is also used for feature extraction to capture important features from the review data. From nine test scenarios, model B3 with 50 nodes in the first hidden layer and a learning rate of 0.1 provided the best performance with an accuracy of 65.85%, precision of 62.27%, recall of 58.61%, and f1-score of 59.71%. Validation using K-Fold Cross Validation shows an average accuracy of 64.17% at k=10. Word analysis with TF-IDF identified dominant words in each category. The B3 model is not yet able to classify reviews perfectly, due to the large and unbalanced dataset, less complex model architecture, and less effective TF-IDF preprocessing. However, this study shows potential for better classification in the future. With optimization, this model can be very useful for e-commerce business owners to gain insight from customer reviews and can help them to identify aspects that will lead to customer satisfaction and trust.
Prediksi Closing Price Saham Harian Berbasis Soft Computing Susilo, Ni`mah Firsta Cahya; Crysdian, Cahyo
Syntax Literate Jurnal Ilmiah Indonesia
Publisher : Syntax Corporation

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36418/syntax-literate.v9i6.15485

Abstract

Prediksi harga saham adalah tugas kompleks yang bergantung pada banyak faktor seperti kondisi politik, ekonomi global, laporan keuangan perusahaan, dan pendapatan. Oleh karena itu, untuk memaksimalkan keuntungan dan meminimalkan kerugian, diperlukan adanya teknik memprediksi nilai saham. Tujuan dari penelitian ini adalah untuk mengetahui pengaruh fitur terhadap perkiraan harga close price saham dan membandingkan kinerja multiple linear regression dan artificial neural network. Uji koefisien korelasi yang telah dilakukan menghasilkan nilai koefisien variabel terbesar dan terkecil dalam penelitian ini adalah 0,9. Algoritma multiple linear regression mengungguli artificial neural network dengan MAPE sebesar 0,900016423%.
Analysis of the Use of Artificial Neural Network Models in Predicting Bitcoin Prices Sahi, Muhammad; Faisal, Muhammad; Arif, Yunifa Miftachul; Crysdian, Cahyo
Applied Information System and Management (AISM) Vol. 6 No. 2 (2023): Applied Information System and Management (AISM)
Publisher : Depart. of Information Systems, FST, UIN Syarif Hidayatullah Jakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15408/aism.v6i2.29648

Abstract

Bitcoin is one of the fastest-growing digital currencies or cryptocurrencies in the world. However, the highly volatile Bitcoin price poses a very extreme risk for traders investing in cryptocurrencies, especially Bitcoin. To anticipate these risks, a prediction system is needed to predict the fluctuations in cryptocurrency prices. Artificial Neural Network (ANN) is a relatively new model discovered and can solve many complex problems because the way it works mimics human nerve cells. ANN has the advantage of being able to describe both linear and non-linear models with a fairly wide range. This research aims to determine the best performance and level of accuracy of the ANN model using the Back-Propagation Neural Network (BPNN) algorithm in predicting Bitcoin prices. This study uses Bitcoin price data for the period 2020 to 2023 taken from the CoinDesk market. The results of this study indicate that the ANN model produces the best performance in the form of four input nodes, 12 hidden nodes, and one output node (4-12-1) with an accuracy rate of around 3.0617175%.