Claim Missing Document
Check
Articles

Found 34 Documents
Search

GOLD PRICESFORECASTING USING TRIPLE EXPONENTIAL METHOD Khairawati, Khairawati; Fuadi, Wahyu; Ramadhansyah, Rizki; Fariadi, Dedi
International Journal of Economic, Business, Accounting, Agriculture Management and Sharia Administration (IJEBAS) Vol. 1 No. 2 (2021): December
Publisher : CV. Radja Publika

Show Abstract | Download Original | Original Source | Check in Google Scholar | Full PDF (376.232 KB) | DOI: 10.54443/ijebas.v1i2.79

Abstract

Governments, organizations, and citizens have taken an interest in gold price fluctuations. Gold price forecasting that is accurate may effectively capture price shift tendencies and reduce the effects of gold market volatility. However, due to the multi-factor and nonlinear nature of the gold market. The triple exponential smoothing strategy is used in this study to predict the rise in a value over time since it can replicate trends and seasonal patterns. according to the gold price swings pattern and seasonal components at the same time To calculate system accuracy, the Mean Absolute Percentage Error is employed (MAPE). With alpha 0.15 and beta 0.85 as parameter values, the triple exponential smoothing (TES) approach achieves an accuracy rate of 86.93 percent and a MAPE of 12.49 percent in this study.
PT. TELEKOMUNIKASI SHARE PRICE PREDICTION ANALYSIS INDONESIA USING THE TRIPLE METHOD EXPONENTIAL Khairawati, Khairawati; Fuadi, Wahyu; Fariadi, Dedi
International Journal of Economic, Business, Accounting, Agriculture Management and Sharia Administration (IJEBAS) Vol. 2 No. 5 (2022): October
Publisher : CV. Radja Publika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.54443/ijebas.v2i5.418

Abstract

The capital market is one of the investment models that is currently growing so rapidly. The amount of interest in investing makes many people who experience losses, due to not understanding the investment risks. This requires technical analysis skills. In this research, we will analyze the prediction of PT. Telekomunikasi Indonesia (TLKM) from 2021 to 2022. The variables used in this study are historical prices ranging from Open, High, Low, and Close prices. The stages used are 184 historical data collection, where the data is taken through Google's financial database and yahoo Finance. Then the calculation process uses the Triple Exponential Smoothing method, the system accuracy process is calculated for the forecast error value using the Mean Absolute Percentage Error (MAPE).
Klasifikasi Kepuasan Masyarakat terhadap Kinerja Petugas Sensus menggunakan K-Nearest Neighbours Auji, Muhammad Fathah; Fuadi, Wahyu; Razi, Ar
Jurnal Algoritma Vol 22 No 2 (2025): Jurnal Algoritma
Publisher : Institut Teknologi Garut

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33364/algoritma/v.22-2.2502

Abstract

This study aims to analyze the level of public satisfaction with the performance of census officers using the K-Nearest Neighbors (KNN) algorithm and the Chebyshev Distance measurement method. In this study, data were collected through field interviews with 239 respondents who were grouped into five satisfaction categories: Very Satisfied, Satisfied, Fairly Satisfied, Dissatisfied, and Very Dissatisfied. The KNN model applied using Python produced an accuracy of 75%, precision of 44%, and recall of 71% with k = 3. The results of the study show that KNN can classify the level of public satisfaction quite well, although the accuracy obtained still shows potential for improvement. This study suggests that further research be conducted using more complex methods to improve classification results.
PENGEMBANGAN MODEL DETEKSI BERITA PALSU PADA PLATFORM BERITA ONLINE MENGGUNAKAN METODE BIDIRECTIONAL ENCODER REPRESENTATIONS FROM TRANSFORMERS (BERT) Dika, Farhan; Fuadi, Wahyu; Afrillia, Yesy
Jurnal Teknologi Terapan and Sains 4.0 Vol 6 No 3 (2025): Jurnal Teknologi Terapan & Sains
Publisher : Universitas Malikussaleh

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29103/tts.v6i3.26167

Abstract

Penyebaran berita palsu pada platform berita online telah menjadi tantangan serius yang mengancam integritas informasi dan stabilitas sosial di era digital. Penelitian ini bertujuan untuk mengembangkan model deteksi berita palsu yang akurat menggunakan metode Bidirectional Encoder Representations from Transformers (BERT) dengan pendekatan fine-tuning pada dataset berita berbahasa Indonesia. Dataset penelitian terdiri dari 248 artikel berita yang dikurasi dari berbagai platform media online, kemudian diseimbangkan menjadi 196 artikel dengan distribusi 50:50 antara berita asli dan berita palsu. Metodologi penelitian mencakup text preprocessing, tokenisasi menggunakan IndoBERT tokenizer dengan panjang maksimal 128 token, dan pembagian data dengan stratified train-test split 80:20. Model IndoBERT di-fine-tune selama 3 epoch dengan konfigurasi batch size 4, learning rate 2e-5, dan gradient accumulation steps 2. Hasil penelitian menunjukkan performa yang sangat baik dengan akurasi 85.0% pada data testing, macro average F1-score 0.8485, dan evaluation loss 0.3669. Model menunjukkan precision 0.9375 dan recall 0.75 untuk deteksi berita palsu, serta precision 0.7917 dan recall 0.95 untuk berita asli. Validasi menggunakan confusion matrix menunjukkan 34 prediksi benar dari 40 sampel testing, dengan karakteristik model yang cenderung konservatif dalam melabeli berita sebagai palsu. Pengujian pada kasus nyata menunjukkan kemampuan model dalam mengidentifikasi indikator linguistik berita palsu seperti kata "SALAH", "PENIPUAN", dan "HOAX" dengan tingkat kepercayaan 65.50%-88.51%. Penelitian ini membuktikan bahwa metode BERT efektif untuk deteksi berita palsu berbahasa Indonesia dan dapat diimplementasikan sebagai sistem moderasi konten otomatis pada platform berita online. Kata kunci: BERT, Deteksi Berita Palsu, Natural Language Processing, Klasifikasi Teks, Deep Learning.