COMPARISON OF ADAPTIVE NEURO-FUZZY AND PARTICLE SWARM OPTIMIZATION BASED NEURAL NETWORK MODELS FOR FINANCIAL TIME SERIES PREDICTION1

Artificial neural networks (ANNs) can be a potential tool for non-linear processes that have unknown relationship and as a result are difficult to fit (Darbellay & Slama 2000). ANNs are non-linear, data-driven and self adaptive approaches as opposed to the above model-based non-linear methods. One of the major application areas of ANNs is forecasting (Zhang, Patuwo, & Hu, 1998). ANN can identify and learn correlated patterns between input data sets and corresponding target values. This technique is In this paper, an attempt has been made to assess the forecasting ability of adaptive neuro-fuzzy inference system (ANFIS) with the traditional feed forward neural network using financial time series data. Also, efforts have been made to examine the performance of particle swarm optimization algorithm for training neural networks. This algorithm is shown to perform well in the current study