Claim Missing Document
Check
Articles

Found 32 Documents
Search

KLASIFIKASI TEXT ULASAN PENGGUNA APLIKASI WONDR BY BNI MENGGUNAKAN ALGORITMA NAIVE BAYES Fitria Ratna Sari; Tukino Tukino; Shofa Shofiah Hilabi; Bayu Priyatna
Djtechno: Jurnal Teknologi Informasi Vol 6, No 2 (2025): Agustus
Publisher : Universitas Dharmawangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.46576/djtechno.v6i2.6819

Abstract

Penelitian ini berfokus pada proses klasifikasi ulasan pengguna aplikasi Wondr by BNI dengan menerapkan algoritma Naïve Bayes. Data yang digunakan berasal dari platform Kaggle, terdiri dari 1.500 data ulasan pengguna aplikasi yang telah melewati tahapan pre-processing seperti cleansing, tokenization, transform cases, stopwords, dan filter tokens. Ulasan tersebut kemudian diberi label secara manual ke dalam kategori label cepat, biasa saja, lambat, dan tidak responsif. Setelah itu label akan di buat otomatis oleh Naïve Bayes. Dataset dibagi menjadi 80:20, lalu di proses menggunakan model klasifikasi berbasis probabilistik Naïve Bayes. Hasil pengujian menunjukkan bahwa algoritma Naïve Bayes mampu mengklasifikasikan ulasan pengguna dengan tingkat akurasi sebesar 95%. Evaluasi model berdasarkan precision, recall, dan f1-score menunjukkan performa klasifikasi yang sangat baik pada setiap kategori ulasan. Visualisasi hasil klasifikasi menggunakan confusion matrix, diagram batang, dan wordcloud memberikan pemahaman lebih mendalam terhadap pola ulasan pengguna. Temuan ini membuktikan bahwa algoritma Naïve Bayes efektif dalam menangani teks tidak terstruktur dan dapat diandalkan untuk mendukung analisis evaluasi layanan digital berbasis umpan balik pengguna.Kata Kunci: Klasifikasi Teks, Naïve Bayes, Ulasan Pengguna, Kaggle, Wondr by BNI
Implementation of The Seasonal Autoregressive Integrated Moving Average Predictive Model on Raw Material Usage Data at PT. Plastik Karawang Flexindo Muhammad Rindra Alfiansyah; Tukino Tukino; Agustia Hananto; Elfina Novalia
International Journal of Engineering, Science and Information Technology Vol 5, No 3 (2025)
Publisher : Malikussaleh University, Aceh, Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52088/ijesty.v5i3.867

Abstract

Fluctuations in raw material utilization in the manufacturing industry significantly impact production process efficiency, operational costs, and supply chain stability. Inaccurate planning and management of raw material inventories can lead to two extreme conditions: excess stock, which increases storage costs and the risk of expiration, or stock shortages, which could halt the production process and reduce productivity. To improve the accuracy of raw material consumption planning, this study applies the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to predict raw material needs periodically based on historical data. The dataset used includes the consumption of Polyethylene (PE), High Density Polyethylene (HDPE), and Polypropylene (PP) from 2019 to 2025. The data is analyzed using a time series forecasting approach to identify trends and seasonal patterns. The SARIMA model is developed and optimized using three methods to search for the best parameters: Grid Search, Random Search, and Bayesian Optimization, to enhance prediction performance. The model's evaluation calculates the Mean Absolute Percentage Error (MAPE) as an accuracy indicator. The evaluation results show that although SARIMA can recognize seasonal patterns in raw material consumption, the prediction accuracy varies, with the best MAPE value being 16% and the highest being 34%. This indicates that external factors, such as market dynamics, government policies, global price fluctuations, and internal variables such as production schedules and customer demand, need to be considered to improve the model's precision. Overall, the application of SARIMA in this context provides a strategic contribution to supply chain management in the manufacturing industry, particularly in anticipating raw material needs, reducing uncertainty, and supporting more efficient and adaptive data-driven decision-making.