Mining Foreign Exchange Rates using Bio-Inspired Neuralnets
Puspanjali Mohapatra1, Soumya Das2, Ashutosh Bhoi3, Tapas Kumar Patra4
1Puspanjali Mohapatra, Assistant Professor, Department of CSE, IIIT, Bhubaneswar (Odisha), India.
2Soumya Das, IIIT, Bhubaneswar (Odisha), India.
3Ashutosh Bhoi, IIIT, Bhubaneswar (Odisha), India.
4Tapas Kumar Patra, Reader I&E CET, Bhubaneswar (Odisha), India.
Manuscript received on 11 June 2013 | Revised Manuscript received on 17 June 2013 | Manuscript Published on 30 June 2013 | PP: 56-62 | Volume-3 Issue-1, June 2013 | Retrieval Number: A0886063113/13©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: To calculate the profit and risk associated with international transactions, currency exchange forecasting is highly desirable. If the forecasting is done accurately then the transaction can give maximum profit.To perform the above task several statistical and machine learning methods have already been proposed by the researchers in the literature. However this paper presents a comparative study between two predominantly used bio-inspired optimization techniques namely particle swarm optimization (PSO) and differential evolution (DE) to forecast the currency exchange rates for one day and one week ahead. For both the algorithms the functional link artificial neural network (FLANN) model is taken into consideration. In the proposed model DE and PSO are used as the evolutionary algorithms for supplementing the optimized value of unknown parameters of the FLANN model. Root mean square error (RMSE) and mean absolute percentage error (MAPE) are considered for performance evaluation of the proposed model. Here JAPANESE YEN(JPY), INDIAN RUPEE(INR), FRENCH FRANC(FRF) to US DOLLAR(USD) datasets are considered as the training and testing datasets.The results of FLANN-DE and FLANN-PSO are analyzed.The simulation results show that FLANN-DE outperforms the FLANN-PSO model regarding the accuracy , convergence speed over different time spans.
Keywords: FLANN, PSO, DE, Currency Exchange Rate Prediction.
Scope of the Article: Data Modelling, Mining and Data Analytics