Session

Computer Science and Information Systems

Description

This paper focuses on the treatment of intelligent systems and their application in the financial area. Types of intelligent systems are numerous, but we will focus on those systems, which based on their ability to learn, are able to predict. The concept of inductive reasoning, how these systems learn and reason inductively, the role and their integration in financial services are some of the concepts that will be addressed. The second and the main part focuses on the application developed in the design of an artificial neural network for financial forecasts. Recognizing the need for better predictive models, not just traditional statistical model, we considered with interest the development of an application that will predict currency exchange rates, USD-ALL, given the time series of real data in years 1995-2012. We test some of the learning algorithms in our system and conclude that one of them is most suitable for this problem. This intelligent system reached to create a relational model of data, on the basis of which is able to output satisfactory results forecast. After the presentation of experimental results, the paper closes with a discussion on possible improvements that could be made in the future.

Keywords:

prediction, time series, currency exchange monetary, neural network

Session Chair

Evelina Bazini

Session Co-Chair

Krenare Pireva

Proceedings Editor

Edmond Hajrizi

ISBN

978-9951-437-24-0

First Page

29

Last Page

36

Location

Durres, Albania

Start Date

1-11-2013 3:15 PM

End Date

1-11-2013 3:30 PM

DOI

10.33107/ubt-ic.2013.55

Download

Share

COinS
 
Nov 1st, 3:15 PM Nov 1st, 3:30 PM

ANN in Financial Prediction

Durres, Albania

This paper focuses on the treatment of intelligent systems and their application in the financial area. Types of intelligent systems are numerous, but we will focus on those systems, which based on their ability to learn, are able to predict. The concept of inductive reasoning, how these systems learn and reason inductively, the role and their integration in financial services are some of the concepts that will be addressed. The second and the main part focuses on the application developed in the design of an artificial neural network for financial forecasts. Recognizing the need for better predictive models, not just traditional statistical model, we considered with interest the development of an application that will predict currency exchange rates, USD-ALL, given the time series of real data in years 1995-2012. We test some of the learning algorithms in our system and conclude that one of them is most suitable for this problem. This intelligent system reached to create a relational model of data, on the basis of which is able to output satisfactory results forecast. After the presentation of experimental results, the paper closes with a discussion on possible improvements that could be made in the future.