Noise Reduction through Bagging on Neural Network Algorithm for Forest Fire Estimates

Forest fires are a common environmental issue and have many negative impacts. Besides to causing damage to the environment, the impact of forest fires is a high cost in dealing with the forest fires themselves and the post-fire recovery process. Estimates of burned areas are important to predicted how strong fire radiation is to the surrounding area, so that resources in dealing with forest fires can be appropriately allocated. In addition, forest fire estimates can provide preliminary information to avoid greater damage. Neural Network is one of the regression and classification methods that can be applied to predict the area of forest fires. However, Neural Network still has weakness when handling noise data. The noise data can affect the results of the experiments performed. To reduce the noise data on Neural Network, in this experiment will be implemented Bagging to get lower error rate on forest fires estimates area. The results of the study will compare the error rates of Neural Network (NN) before and after Bagging implementation (NN+BG).