Optimization of the Use of Artificial Neural Network Models for Accuracy Data Measurement Palm Oil Production Prediction Rate

Predictions are necessary in modern times if the amount of production, profit and loss analysis, and development must be calculated on a specific scale. The average prediction only determines a certain period using data series, so it will be very helpful on certain production scales, especially palm oil production as a commodity with a very large amount of investment. Palm oil production is Indonesia's production potential, which requires a prediction pattern for a certain period. The autoregressive integrated moving average (ARIMA) model is used to calculate production rates by performing predictive data analysis of palm oil production development over a period of time. To prove the level of foresight in the prediction process, the next step is to calculate the level of foresight of the prediction process error through the Mean Square Error (MSE) model. To get the accuracy value, the Artificial Neural Network (ANN) algorithm approach is carried out using the MATLAB tool. There are several ways to test the level of accuracy; the results are compared with the calculation results using MS Excel as well as the Statistical Product and Service Solution (SPSS) tool. The calculated accuracy of SPSS (0.015) and MS Excel (0.0201) are known, while the ANN algorithm is more observant with a value of 0.0055.