Electrical Power Sold Prediction using the MLR (Multiple Linear Regression) methods optimized by FFNN (Feed Forward Neural Network)

A decent standard of living is a standard that is adequate for each person for himself and his family, relating to basic needs, and the sustainable improvement of living conditions. The poverty line is the minimum income level to obtain a decent standard of living. On the other hand, HDI (Human Development Index) is an aggregate indicator that provides a complete picture of the quality of life. Of course, Prosperous Families have met a decent standard of living, and vice versa for Poor Families (contributing to the percentage of poor people or % PP). In this case, the amount of electricity consumption and NEC have a positive effect on HDI, and vice versa on % PP. Multiple linear regression is a technique that can be used to test these assumptions through model estimation. This research measures the influence of HDI, NEC and % PP on EPS. Analysis of the desired effect was carried out using the MLR (Multiple Linear Regression) method. The MLR model obtained was then optimized using FFNN (Feed Forward Neural Network) to predict EPS. The results of this study showed that the assumption has proven, where HDI $(\beta_{1}=0$. 7853) and NEC $(\beta_{3}=0$. 7849) have a positive influence, and vice versa % PP has a negative influence $(\beta_{2}=-0.0783)$ by using the OLS estimator. The results EPS prediction have shown that the optimized MLR model provides much better performance (% $R_{square}=$149.26%) compared to the MLR model without optimization (% $R_{square}=$114.99%) in terms of predictive activity. Performance improvement that occurs was equal to 29.80% on $R_{square}$, 44.98% on RMSE, and 73.76% on MAPE.