Epidemiological time series data are modeled efficiently by the proposed NFGF.
The key success of the NFGF comes from the recurrent nodes and the GA optimizer applied to the fuzzy rule structure. While addition of recurrent nodes increases number of modifiable parameters then GA mitigate this computational burden by reducing fuzzy rules.
The harmonious combination of GA and the recurrent node extension brings an optimal fusion of soft computing techniques that is able to cope with non-linear and temporal relations in epidemiological time series.
Results of experiments in time series modeling shows superior performance and flexibility of NFGF compared to ANFIS and FFNN.