This research discusses the forecasting of the median value of owner occupied homes (MEDV) using all the other continuous variables available in the Boston dataset. The Boston dataset is a collection of data about housing values in the suburbs of Boston. The used method is Feed Forward Neural Network (FFNN) and the multiple linear regression method as a comparison. The result of the research indicates that the FFNN method is better than multiple linear regression in forecasting the median value of owner occupied homes (MEDV) using all the other continuous variables available in the Boston dataset. It is proven that the MSE and MAPE value of using the FFNN method is 15.7518 and 0.14563, whereas the value of multiple linear regression is 31.2630 and 0.21040. Based on this result, the research can be concluded that the FFNN method has the smaller MSE value, the result of the forecasting is more accurate.
Copyrights © 2023