Stevens response

A machine learning algorithim is an a statisticle technique which allows your computer/device on hand to learn the best method of helping solve a problem. In the context of this study, the researchers used the random forest machine learning technique in which they incorporate different weighted geospatial covariates to help predict whether or not people live within a certain area. This is different from a regular statistical method because it produces an unequal distribution of population where as classical statistical models will have a set, uniform distribution. Using random forest with covariates, the algorithm will help the device understand which areas are likely to have people who live in it. The researchers used amples amounts of geospatial datasets. For instance, they relied heavily on night time lights and distance to hospitals. In fact the researchers noted that the distance to hospitals had a relative significant impact on the model for Kenya. In the case of this study, the covariates represented the geospatial aspects of the entirity of the country down to specific details. Big data is significant for machine learning because the more data an algorithim has, the more a machine will be able to learn about the topic because of stronger correlations. This relates to the importants of highly accurate descriptions where each person lives. This is important because the descriptions are the indicators to the machine learning algorithms of where people live. So if the indicators are inaccurate, the predictions will likely also be inaccurate. Having highly relevant/accurate data, in the context of my LMIC, is extremely important because there are not many census data taken. So, the geospatial covariates need to be as accurate as possible to compensate.