Forecasting The Exchange Rate Series With Ann: The Case Of Turkey

Abstract

Zaman serilerindeki hem do g rusal, hem de do g rusal olmayan yapiyi Yapay Sinir A g lariyla (YSA) modellemek mumkun oldu g undan, YSA yonteminin do g rusal olmayan yapiya sahip kaotik serilerin modellenmesinde de kullanimi uygun olacaktir. Bu nedenle, yapilan cali s mada, yuksek dalgalanma gosteren Turkiye TL/US dolar doviz kuru zaman serisinin modellenmesinde YSA yontemi kullanilmi s tir. Elde edilen sonuclar gore, YSA yonteminin mevsimsel ARIMA ve ARCH gibi modellerden daha iyi ongoruler uretti g i gorulmu s tur. Ek olarak, Turkiye doviz kuru zaman serisinin cozumlenmesinde, YSA yontemi kullaniminin detaylari da verilmi s tir. Anahtar Kelimeler: Aktivasyon fonksiyonu, ARIMA, ARCH, Doviz kuru, Kaotik seriler, Ongoru, Yapay sinir a g lari, Zaman serileri Jel Siniflamasi: C220, C450, C530, F310, G170 ABSTRACT As it is possible to model both linear and nonlinear structures in time series by using Artificial Neural Network (ANN), it is suitable to apply this method to the chaotic series having nonlinear component. Therefore, in this study, we propose to employ ANN method for high volatility Turkish TL/US dollar exchange rate series and the results show that ANN method has the best forecasting accuracy with respect to time series models, such as seasonal ARIMA and ARCH models. The suggestions about the details of the usage of ANN method are also made for the exchange rate of Turkey. Keywords : Activation function, ARIMA, ARCH, Artificial neural network, Chaotic series, Exchange rate, Forecasting, Time series Jel Classification: C220, C450, C530, F310, G170

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 31 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.