This work presents new algorithms for feature selection. The main propose is introduce the Relief algorithm to obtain an importance classification of the attributes to find which are less important. By removing the worst, an inducer will give us the performance to choose the best subset of data.