<![CDATA[Growth and development of children at an early age affect the child's personal ability in the future. Every child is unique, so growth and growth are not the same. Slow growth and development are often considered normal. Deviation of late child growth is known to result in long-term and difficult to repair. Based on these problems, this research was conducted by using the Extreme Learning Machine (ELM) method for the classification of child growth deviations. ELM method consists of training process as system learning and testing to obtain the result of classification. The parameters test are test of ratio of training data and test data, testing the influence of number of hidden neurons over time, and comparative test of activation function. Accuracy calculation is done by using confusion matrix to know the accuracy of system work in each class. The result of parameter test shows that the ratio of training data and test data with ratio 70:30, the number of hidden neurons as many as 10 units, and the binary activation function is the parameter with the best accuracy value. The comparison of the result of the classification of child growth deviation with the help of psychologist shows that the system produces poor accuracy. This can be due to the small and unbalanced data used for the research.]]>
Copyrights © 2018
