Shallots are one of the leading commodities that strengthen national food security. From 2013 to 2018 the development of shallots production had increased. Except in 2015 the production of shallots decreased by 0.39 percent compared to 2014. Prediction of shallots prices is needed in order to maintain price stability for supporting food security, economic stability, and trade. In predicting the price of shallots commodities, statistical modeling is carried out using parametric and nonparametric time series approaches. However, in this research the parametric approach did not meet the assumption of white noise. Therefore, the nonparametric approach of kernel estimator and Fourier series estimator was used with correlated error. Nonparametric approach is used because it has a flexible form and alternative solutions if the parametric approach does not meet the assumptions. The result was the best model to predict of shallots prices in Indonesia was modeled based on the nonparametric approaches with kernel estimator. The model met goodness criteria like the small MSE value is 757.7224 and the big determination coefficient is 99.95%. The goodness criteria for kernel estimator is better than Fourier series estimator. The kernel estimator has good performance to predict the price of shallots with small MAPE value is 1.088%.