Setiawan, Dwi Novi (2018) Prediksi Nilai Tukar Rupiah Terhadap Dolar Amerika Dengan Menggunakan Metode Algoritme Genetika - Backpropagation. Sarjana thesis, Universitas Brawijaya.
Abstract
Nilai tukar (kurs) merupakan nilai mata uang suatu negara yang dinyatakan ke dalam bentuk mata uang negara lain. Nilai tukar berperan penting dalam perdagangan internasional. Untuk menjaga kestabilan nilai tukar rupiah, maka pemerintah perlu memberlakukan kebijakan yang tepat. Oleh karena itu diperlukan sebuah algoritme prediksi yang mampu mengenali pola dari perubahan nilai tukar rupiah. Metode Backpropagation merupakan salah satu metode yang mampu mengenali pola pada data time series, sedangkan Algoritme Genetika merupakan metode yang mampu melakukan eksplorasi solusi yang lebih luas bagi Backpropagation. Pada Algoritme Genetika, bobot Backpropagation direpresentasikan dalam bentuk real-code. Implementasi Algoritme Genetika – Backpropagation memiliki tahapan inisialisasi populasi, crossover, mutasi, pelatihan setiap individu menggunakan Backpropagation, evaluasi, dan seleksi. Dari pengujian yang telah dilakukan didapatkan parameter yang paling optimum bagi Algoritme Genetika – Backpropagation jumlah generasi 90, ukuran populasi sebesar 20, crossover rate 0,1, mutation rate 0,9, jumlah neuron pada lapisan tersembunyi 13, learning rate 1 dan jumlah iterasi pelatihan Backpropagation sebesar 500. Hasil pengujian yang telah dilakukan didapatkan nilai MAPE terbaik sebesar 1,575318 dan rata-rata MAPE sebesar 1,741747. Algoritme juga mampu melakukan validasi dengan MAPE terbaik sebesar 1.0004917 dan rata-rata MAPE sebesar 1,077603. Algoritme Genetika mampu meningkatkan akurasi Backpropagation hingga 5,947759%.
English Abstract
The exchange rate is the value of the currency of a country which is expressed in the form of currency of another country. Exchange rate has an important role in international trade. To maintain the stability of the rupiah exchange rate, the government needs to enact the right policy. Therefore, a prediction algorithm that is able to recognize the pattern of exchange rate changes is needed. Backpropagation is one of method that is able to recognize patterns in time series data, while Genetic Algorithm is one of the capable method to exploring wider solutions for Backpropagation. In the Genetic Algorithm, the weight of Backpropagation is represented in real-code. Implementation of Genetic Algorithm - Backpropagation has initialization phase of population, crossover, mutation, individual training using Backpropagation, evaluation, and selection. The most optimum parameters for Genetic Algorithm - Backpropagation are in 90 generation, 20 population size, 0.1 crossover rate, 0.9 mutation rate, number of neurons in hidden layer 13, learning rate 1 and number of iteration of Backpropagation training were 500. The results of the tests that have been done got the best MAPE value of 1.575318 and the average MAPE of 1.741747. The algorithm is also capable of performing validation with the best MAPE of 1.0004917 and the average MAPE of 1.077603. Genetic Algorithm can increase the accuracy of Backpropagation up to 5.947759%.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2018/364/051805145 |
| Uncontrolled Keywords: | Nilai Tukar, Kurs, Algoritme Genetika, Jaringan Syaraf Tiruan, Backpropagation, Algoritme Genetika - Backpropagation Exchange Rate, Genetic Algorithm, Artificial Neural Network, Backpropagation, Genetic Algorithm – Backpropagation |
| Subjects: | 000 Computer science, information and general works > 006 Special computer methods > 006.3 Artificial intelligence > 006.31 Machine learning |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 31 Jan 2019 02:36 |
| Last Modified: | 18 Oct 2021 01:22 |
| URI: | http://repository.ub.ac.id/id/eprint/13443 |
Preview |
Text
Dwi Novi Setiawan.pdf Download (3MB) | Preview |
Actions (login required)
 |
View Item |