Artificial Neural Network Model for Estimating the Poor Population in Indonesia as an Effort to Alleviate Poverty

Abstrak

Forecasting poverty levels in Indonesia is an activity that is considered likely to help various parties, such as helping the government in planning more effective and efficient poverty alleviation programs. In the research, forecasting of poverty levels in Indonesia was carried out using the artificial network method backpropagation. The aim of this study is to model and predict poverty levels using artificial neural network models backpropagation, as well as to determine the level of accuracy of the forecasting results produced by this method. This research is applied research. The data used is annual poverty data in Indonesia from 2017-2021. The data is then divided into two parts, namely training data and test data. The research results show that the best artificial neural network model is BP (7,7,2) with 7 neuron at the input layer, 7 neuron on the hidden layer, and 2 neuron at the output layer. The accuracy level of this model is good with a MAPE value of 0.07633%. Forecasting results for the next period show that the highest number of poor people is in the province of East Java with a value of 3604.1698 thousand people in the 1^st semester (March) of 2022 and experienced an increase in the 2^nd semester (September) of 2022 with a value of 3698.822 thosand people.