Claim Missing Document
Check
Articles

Found 4 Documents
Search

Experimenting Diabetic Retinopathy Classification Using Retinal Images Muhammad Fermi Pasha; Mark Dhruba Sikder; Asif Rana; Maya Silvi Lidya; Ronsen Purba; Rahmat Budiarto
Data Science: Journal of Computing and Applied Informatics Vol. 5 No. 1 (2021): Data Science: Journal of Computing and Applied Informatics (JoCAI)
Publisher : Talenta Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32734/jocai.v5.i1-5232

Abstract

Along with many complications, diabetic patients have a high chance to suffer from critical level vision loss and in worst case permanent blindness due to Diabetic Retinopathy (DR). Detecting DR in the early stages is a challenge, since it has no visual indication of this disease in its preliminary stage, thus becomes an important task to accomplish in the health sector. Currently, there have been many proposed DR classifier models but there is a lot of room to improve in terms of efficiency and accuracy. Despite having strong computational power, current deep learning algorithm is not able to gain the trust of the medical experts in classifying DR. In this work, we investigate the possibility of classifying DR using deep learning with Convolutional Neural Network (CNN). We implement preprocessing combined with InceptionV3 and VGG16 models. Experimental results show that InceptionV3 outperforms VGG16. InceptionV3 model achieves an average training accuracy of 73.5 % with a validation accuracy of 68.7%. VGG16 model achieves an average training accuracy of 66.4% with a validation accuracy of 63.13%. The highest training accuracy for InceptionV3 and VGG16 is 79% and 81.2%, respectively. Overall, we achieve an accuracy of 66.6% on 52 images from 3 different classes.
Pemanfaatan Analisis Sentimen dari Ulasan Produk di Youtube untuk Pengembangan Produk Baru Limbong, Ricky Paian; Ronsen Purba; Muhammad Fermi Pasha
Syntax Literate Jurnal Ilmiah Indonesia
Publisher : Syntax Corporation

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36418/syntax-literate.v9i7.13568

Abstract

Pengembangan produk yang sukses memerlukan pemahaman tentang kebutuhan dan preferensi pelanggan. Analisis sentimen telah muncul sebagai alat yang dapat mengumpulkan pendapat dari pelanggan dalam mengembangkan yang lebih baik. Penelitian ini bertujuan untuk mengeksplorasi pemanfaatan analisis sentimen dari ulasan produk di YouTube dalam rangka pengembangan produk baru. Dengan menganalisis konten yang dibuat oleh pengguna, penelitian ini bertujuan untuk menghasilkan informasi berupa prioritas fitur produk. Metode penelitian meliputi pengumpulan dan prapemrosesan data ulasan produk dari platform YouTube, dengan menerapkan teknik pemrosesan teks seperti case folding, penghilangan kata yang tidak relevan, tokenisasi, dan stemming. Analisis sentimen dilakukan menggunakan metode Support Vector Machine (SVM) untuk mengklasifikasikan sentimen yang diekspresikan dalam ulasan tersebut. model yang telah dilatih kemudian digunakan untuk memprediksi dan memberi label sentimen pada ulasan produk baru. Temuan penelitian ini menunjukkan bahwa analisis sentimen dapat membantu proses pengembangan produk baru dengan memperhatikan prioritas fitur produk yang memiliki kekurangan. Pendekatan ini memungkinkan perusahaan untuk memahami kebutuhan pelanggan, membuat keputusan yang tepat dalam memberikan fokus untuk peningkatan fitur produk untuk perilisan selanjutnya. Integrasi analisis sentimen dalam proses pengembangan produk baru dapat memanfaatkan opini konsumen untuk merilis produk yang lebih baik.
New Approach: Customer Segmentation using RFM Model and Demand Classification Fewie Rusly; Ronsen Purba; Muhammad Fermi Pasha
Jurnal Teknologi dan Manajemen Informatika Vol. 11 No. 2 (2025): Desember 2025
Publisher : Universitas Merdeka Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26905/jtmi.v11i2.16208

Abstract

This research introduces an integrated data mining framework that combines RFM (Recency, Frequency, Monetary) analysis with demand pattern classification—encompassing Smooth, Erratic, Intermittent, and Lumpy categories—to refine customer segmentation strategies. While RFM effectively captures transactional behavior, its scope remains insufficient as it overlooks demand variability and intermittency, which critically influence purchasing dynamics and inventory planning. By incorporating demand classification, this model addresses behavioral dimensions beyond conventional transactional metrics, thereby enhancing segmentation precision and strategic relevance. Customer clustering employs the K-Means algorithm, with cluster optimization validated through Elbow Method and Silhouette Index analyses, yielding five distinct segments: Ideal, Interest, Improve, Inconsistent, and Inactive. Subsequently, Customer Lifetime Value (CLV) is computed by weighting RFM and demand parameters via Analytic Hierarchy Process (AHP), with Consistency Index and Consistency Ratio assessments ensuring methodological rigor. Results are synthesized within an interactive dashboard, facilitating data-driven decision-making in retention strategies, inventory optimization, profitability enhancement, and sustainable business development.
Implementation of the CNN-LSTM Hybrid Model in Predicting Bitcoin Price Fluctuations Candra Wibowo; Ronsen Purba; Muhammad Fermi Pasha
Jurnal Teknologi dan Manajemen Informatika Vol. 11 No. 2 (2025): Desember 2025
Publisher : Universitas Merdeka Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26905/jtmi.v11i2.16239

Abstract

Digital financial systems of today face formidable obstacles from the extreme price volatility and unpredictability of Bitcoin. Data cleaning, Min-Max normalization, and sequence creation with a sliding window were performed on the daily BTC-USD historical data received from Yahoo Finance from 2020 to 2024 before implementing a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model in this study. The CNN layers are responsible for extracting local patterns with a limited time horizon, whereas the LSTM layers are responsible for capturing the time series' long-term relationships. The experimental findings show that the CNN-LSTM model outperforms the CNN and LSTM in terms of predictive ability, with an RMSE of 2,202.717, an MAE of 1,553.202, and a MAPE of 2.244%, which translates to an accuracy of about 97.756%. These results provide useful information for adaptive trading techniques and digital asset risk management based on artificial intelligence, and they prove that the hybrid method is successful in dealing with complicated, non-linear, and unpredictable trends in the cryptocurrency market.