**3.3. Generation of the starting population of genetic individuals**

There must be a starting population so that the evolution algorithm can act, through the application of the selection, crossover and evolution operators. For this, aiming at the variability of individuals and consequent improvement on the precision of results, we adopted the *Ramped Half-and-Half* [24] technique.

This technique selects, initially, a random value to be the maximum depth of the tree to be generated. Next, the method for generation of the new tree is selected. *Ramped Half-and-Half* uses two generation methods, where each one generates half of the population. They are described below:

