Rainfall and Wave Height Prediction in Semarang City Using Vector Autoregressive Neural Network (VAR-NN) Methods

In the past few months in Semarang City, high rainfall and high waves give negative impact because the activities of fishermen to be stopped due to the waves of water. Another disadvantage is ROB flood. This is quite detrimental in addition to disrupting transportation and economic activities, but also causing damage to both government-owned assets and personal assets. This damage is caused by flood water which is mixed with sea water containing salt, so that it can cause corrosion. Vector of Autoregressive Neural Network (VAR-NN) is a development of the Autoregressive Vector (VAR) method using the Neural Network (NN) algorithm. The VAR method does not need to distinguish between endogenous and exogenous variables, its means that all variables used in the VAR model are used as endogenous variables. The VAR model can be used to explain and make predictions from endogenous variables from past data on these variables. VAR is developed using the Neural Network (NN) algorithm, where the basic principle of NN itself is an information processing algorithm that resembles the workings of the human brain, which uses a number of neurons to perform simple tasks. VAR-NN is often used in the economic, financial fields and believed to be able to make predictions of high rainfall and wave height. In this study using VAR-NN with FeedForward algorithm or called Backpropagatian. The results of rainfall prediction and wave height in Semarang city using this method have a Mean Square Error of 12,325 with the VAR-NN model (3, 2, 4).