Currently, there is an initial prediction of top McMurray based on nothing more than the thickness of the McMurray in the nearest neighbor well and where that would place the top McMurray based on the known base McMurray. This single feature resulted in a drop in the root mean square error from 90 to 40.
The next step would be to make the same feature / prediction based on the average N wells within a M distance.
Currently, there is an initial prediction of top McMurray based on nothing more than the thickness of the McMurray in the nearest neighbor well and where that would place the top McMurray based on the known base McMurray. This single feature resulted in a drop in the root mean square error from 90 to 40.
The next step would be to make the same feature / prediction based on the average N wells within a M distance.