Laporkan Masalah

Penerapan dan Analisis Kompoaratif Metode Fuzzy Time Series Chen dan Hybrid Quantile Fuzzy Time Series-Genetic Algorithm untuk Peramalan Harga Cryptocurrency Bitcoin

Muhammad Fakhri Zain Pribadi, Dr. Irwan Endrayanto Aluicius, S.Si., M.Sc

2026 | Skripsi | STATISTIKA

Peramalan harga cryptocurrency Bitcoin menjadi salah satu permasalahan yang penting untuk dikaji karena Bitcoin memiliki karakteristik harga yang sangat fluktuatif dan sulit diprediksi. Penelitian ini dilakukan untuk menerapkan dan membandingkan performa metode Fuzzy Time Series Chen dan Hybrid Quantile Fuzzy Time Series–Genetic Algorithm dalam meramalkan harga cryptocurrency Bitcoin. Data yang digunakan merupakan data harga penutupan harian dan volume transaksi Bitcoin yang diperoleh dari Yahoo Finance pada periode 23 April 2025 hingga 22 April 2026. Metode Fuzzy Time Series Chen digunakan sebagai model dasar dengan pendekatan univariat berdasarkan variabel harga penutupan, sedangkan metode Hybrid Quantile Fuzzy Time Series–Genetic Algorithm menggunakan pendekatan bivariat dengan melibatkan variabel return harga dan return volume. Performa peramalan diukur menggunakan Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), dan Mean Absolute Percentage Error (MAPE). Hasil akhir menunjukkan bahwa metode Fuzzy Time Series Chen pada data pengujian menghasilkan RMSE sebesar 4217,38, MAE sebesar 3730,50, dan MAPE sebesar 5,32%. Sementara itu, metode Hybrid Quantile Fuzzy Time Series–Genetic Algorithm memperoleh kombinasi parameter terbaik (k,h,m)=(10,10,3) dengan RMSE validasi sebesar 2361,01, serta menghasilkan RMSE sebesar 1667,44, MAE sebesar 1305,47, dan MAPE sebesar 1,87% pada data pengujian. Oleh karena itu, pada studi kasus yang dilakukan dalam penelitian ini, metode Hybrid Quantile Fuzzy Time Series–Genetic Algorithm mampu menghasilkan performa peramalan yang lebih baik dibandingkan metode Fuzzy Time Series Chen dalam meramalkan harga cryptocurrency Bitcoin.

Forecasting the price of Bitcoin cryptocurrency is an important problem to study because Bitcoin has highly fluctuating price characteristics and is difficult to predict. This research was conducted to apply and compare the performance of Chen's Fuzzy Time Series and Hybrid Quantile Fuzzy Time Series-Genetic Algorithm in forecasting Bitcoin cryptocurrency prices. The data used in this research consist of daily closing price and trading volume data of Bitcoin obtained from Yahoo Finance for the period from April 23, 2025 to April 22, 2026. Chen's Fuzzy Time Series method is used as a baseline model with a univariate approach based on the closing price variable, while the Hybrid Quantile Fuzzy Time Series-Genetic Algorithm method uses a bivariate approach involving price return and volume return variables. Forecasting performance is measured using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The final results show that Chen's Fuzzy Time Series method on the testing data produces an RMSE of 4217.38, an MAE of 3730.50, and a MAPE of 5.32%. Meanwhile, the Hybrid Quantile Fuzzy Time Series-Genetic Algorithm method obtains the best parameter combination of (k,h,m)=(10,10,3) with a validation RMSE of 2361.01, and produces an RMSE of 1667.44, an MAE of 1305.47, and a MAPE of 1.87% on the testing data. Therefore, in the case study conducted in this research, the Hybrid Quantile Fuzzy Time Series-Genetic Algorithm method is able to produce better forecasting performance than Chen's Fuzzy Time Series method in forecasting Bitcoin cryptocurrency prices.

Kata Kunci : Peramalan, Fuzzy Time Series, Fuzzy Time Series Chen, Quantile Fuzzy Time Series, Genetic Algorithm, Bitcoin, Cryptocurrency.

  1. S1-2026-496414-abstract.pdf  
  2. S1-2026-496414-bibliography.pdf  
  3. S1-2026-496414-tableofcontent.pdf  
  4. S1-2026-496414-title.pdf