There are available several linear time-series forecasting models in the literature. In which important and most comman technique for analysis of univariate time-series is Autoregressive integrated moving average (ARIMA) methodology (Box et al., 2007). Sometimes addition of the other exogenous variables increases the prediction accuracy of ARIMA model (ARIMAX). For this aspect, we applied different p and q order ARIMAX model for five nutrient combinations of nitrogen content, which is further, developed by including organic carbons an input (exogenous) variable. Among the linear models, the ARIMAX model performed better as compare to ARIMA model. However, the performance of machine intelligence techniques like Hybrid of linear and nonlinear model is better as compared to linear time series models. The variations in nitrogen uptake for rice crop data for all treatments are large. This could be the reason that Hybrid of linear and nonlinear model found heterogeneous trend in the data set and performed well as compare to ARIMA and ARIMAX. Further the comparision of the forecasted values by hybrid model of different treatments ( control, 100% NPK, 100% NPK+ Zn, 100% NPK+ FYM, 50% N+100% PK+ GM) result has been concluded that 100 % NPK+FYM treatment gave highest forecasted value in comparison to other treatments.