<![CDATA[Linear ordinary differential equations are a type of differential equation that is generally easy to solve analytically when the function on a partial integral has a simple form. However, when the function is a difficult function, it requires other methods such as numerical methods and methods adapted from neural networks because analytical methods can only be used when the problem has a simple geometric interpretation. This study involves the Euler method followed by error estimation using neural networks and the Runge-Kutta Orde-4 method as a comparison. The comparison was carried out by solving four equations which were then analyzed for the results and errors in each method based on the graphs generated and the MAPE criteria. The results of the study based on graphs show that the error generated by the method with error estimation using neural networks is more stable than the 4th Order Runge-Kutta method. In addition, based on the results of calculations with the MAPE criteria, the error estimation method using neural networks produces a very high level of accuracy in the category, while the 4th Order Runge-Kutta method produces a level of accuracy in two categories, namely the very high and reasonable categories]]>
Copyrights © 2025
