<![CDATA[This study aims to compare performance of central government cash planning model using ANFIS (Adaptive Neuro-Fuzzy Inference System) and ARIMA (Autoregressive Integrated Moving Average) methods. The data used is the balance of realization unscheduled expenditure in the Province of DKI Jakarta within period 2015 to 2019 with a total of 48 datasets used as training data and 12 datasets for data checking. This research was conducted by comparing the performance of models in prediction using the RMSE (root mean square error) value generated by each model as the basis for evaluation. Data processing is assisted by the Eviews application to generate ARIMA models and Matlab applications for ANFIS models. The conclusion is that the ARIMA model has a better performance compared to ANFIS with a data timeframe of 60 datasets which results in a smaller RMSE value. The implication of this research is that the use of the ARIMA method can be used effectively on small datasets (short term) and the use of the ANFIS and ARIMA methods to predict budget realization balances can be applied to the Directorate General of Treasury as the manager of the state treasury to support data-based policies.]]>
