Once the model is trained, it can be used to label new data. Each of the individual trees generates a label. The final prediction is determined by voting these individual predictions, as shown:

Note that in the preceding diagram, m trees are trained, which is represented by C₁ to C_m. That is Trees = {C₁,..,C_m}

Each of the trees generates a prediction that is represented by a set:

Individual predictions = P= {P₁,..., P_m}

The final prediction is represented by P_f. It is determined by the majority of the individual predictions. The mode function can be used to find the majority decision (mode is the number that repeats most often and is in the majority). The individual prediction and the final prediction are linked, as follows:

P_f = mode (P)