Consider the ratings matrix in Figure 2.1 (an extension of the sample matrix in Figure 1.1). We want to find User C’s prediction for Equilib- rium (pC,e) with the following configuration:
• Pearson correlation.
• Neighborhood size of 2.
• Weighted average with mean offset (Equation (2.6)).
C’s mean rating is 3.667. There are only two users who have rated Equilibrium, and therefore only two candidate users for the neighbor- hood: A and D. s(C,A) = 0.832 and s(C,D) = −0.515. The prediction pC,e is therefore computed as follows: