Elpiji cylinder (Liquefied petroleum gas) are basic life needs of the general public. Unfortunately for customers, unstable elpiji price in retailer. The Indonesian Government plays an important role for the Liquefied petroleum gas industry to give elpiji price stable for end user. This work use artificial neural network and backpropagation for prediction of elpiji price. Total of 1096 records collected from 2015 until 2017 were fed into the neural network models with nine variable for input data. There are inflation, elpiji Allocation, elpiji prices previous, the poor society (the poor, very poor, the near poor), and date (year, month, day). This data were used to evaluate prediction accuracy, and the price prediction results were found to be more accurate than those made by a method using only eight input variable. Root Mean Square Error (RMSE) nine variable 0.030959131, RMSE variable just 0.199884634, and the test result 0.121417236. The presented results were proved that this model can be used with good accuracy for the prediction elpiji price.

Keywords : Artificial Neural Network, Backpropagation, Elpiji, Prediction, Root Mean Square Error