Net Asset Value Prediction Using Extreme Learning Machine with Dolphin Swarm Algorithm

Net Asset Value prediction of different mutual funds is highly essential for different investors. But prediction of such financial time series data is very difficult task as NAV data are nonlinear in nature. Various researchers have used feed forward neural network models involving gradient descent learning in NAV prediction. But these models have certain limitations of local optima and slow convergence. In this work, Extreme Learning Machine (ELM) with Dolphin Swarm Algorithm (DSA-ELM) is taken as the prediction model. Dolphin Swarm Algorithm (DSA), a swarm based optimization algorithm based on the living habits of dolphins is considered for the optimization of ELM parameters. This DSA-ELM model is used for the NAV prediction of two of the Indian mutual funds and the performance is compared with the standard ELM. From the results, it is observed that the prediction performance of DSA-ELM is better as compared to standard ELM.