Using the models to predict themselves, the standard deviations for each of these tests ranges from 3.9 dB to 4.5 dB and the mean was very close to 0. For example, the model of cell site B was able to predict the actual measured values for cell site B with a standard deviation of 3.87 dB and a mean of 0.06 dB.

Using cell site A to predict cell site D, which are located at the same place, had a standard deviation of 6.78 dB and a mean of 0.75 dB was obtained. This is encouraging results because the predictions of the four unknown sites should be around these values as well if the McAllister Dodger model is accurate.

As previously stated, this model is dependent on the classification of the 8 regions discussed in the Terrain Classification section. The McAllister Dodger model could become more generalized by just classifying strictly the indoor and outdoor regions, at a cost of a higher standard deviation. However, the regions do not have to be exact so some approximation can be used to determine the outdoor regions.

If it is too uneconomical to measure data values for the received power at locations around the base station, the Path Loss Exponent model would most likely be more accurate than the McAllister Dodger model. The McAllister Dodger model becomes more accurate proportional to the amount of data locations used to train the correction factors.