By studying the standard deviation between the resulting predicted values of our multivariate linear regression and the calculated values, the map of standard deviations for each district shown in Fig 1. was produced. The highest overestimation of the predicted neonatal mortality was found in some the districts forming part of the capital Province, Phom Penh: Dangkao, Saensokh, and Ruessei Kaev. This districts are in dark blue in the map. The districts with the highest underestimation of the predicted neonatal mortality will be studied now. In general, this districts are those that had the highest values of the dependent variable and which the model wasn't able to predict a high enough value. There are 5 of this districts, four of them situated along the central longitude of the country from the center to the south and one of them situated to the east. Taking the example of the district with the highest underestimation, Krong Kampong Chhnang, the small district situated at the center of the country, the measured value of the neonatal mortality was of 45 cases per 1000 infants, but the predicted value was of only 25 cases per 1000. It is interesting to remark that most of the districts that had either a high overestimation or underestimation have a very small surface area.