Seed Laboratory BPTP East Java is one of provincial government work units that was appointed as the technical implementer to conduct a study in seed growth. Currently this facility is developing automatic watering device based on soil humidity sensor, but the device cannot predict the volume of water needed in order to keep the soil moisture. With the help of humidity sensor on device and expert's knowledge, the observations dataset of soil moisture to the waters volume requirement are obtained. This study was conducted to implements linear regression method so that the device can perform predictions based on dataset patterns as an equation. The accuracy of prediction results with this method is measured by the coefficient of determination. The coefficient of determination can be decreasing due to the arising of observation outliers because Inaccuracy of observation results. The solution from this study is using evolution strategies with information criteria as comparison for detecting observation outliers to eliminated. After eliminating 6 observations outliers were detected by evolution strategiess in this study, shows increase in the coefficient of determination from 0.9673 to 0.9935
Copyrights © 2019