We refer to these vertical discrepancies as absolute deviations (Figure 3.5). For the best-fitting line, we might try to minimize the sum of there absolute deviations, Leading to the model depicted in Figure 3.5. Although success may be achieved in minimizing the sum of the absolute deviation from individual point may be quite large. For example, consider point D in Figure 3.5. If the modeler has confidence in the fitted line near the point.