Wavelets Based Artificial Neural Network Technique for Forecasting Agricultural Prices

It has been observed that most of the agricultural time series data in general and price particular are non-linear, non-stationary, non-normal heteroscedastic nature. Therefore, application usual linear nonlinear parametric models like Autoregressive integrated moving average (ARIMA), Generalized autoregressive conditional (GARCH) their component fail to capture variability present series. is also very difficult extract actual signal from noisy observations. In this regard, nonparametric wavelet technique advantage pre-processing signal. Optimizing level decomposition choosing appropriate filter can represent with high chaotic nature sophisticated structure more effectively. The describe useful pattern both global as well local perspective. decomposed components be modeled using Machine Learning techniques Artificial Neural Network (ANN) result wavelet-based hybrid eventually, inverse transform carried out obtain prediction original above algorithm applied for modeling monthly modal wholesale tomato Burdwan market, West Bengal, India. Haar D4 filters have two levels i.e. 3 6. accuracy model compared empirically ARIMA, GARCH ANN it outperformed other models.