Link Prediction SVM CrossValidation, problem with low F scores

Question

I\'m trying to predict new edges in a Graph I converted it into a dataFrame with columns of \'source\' \'destination\' \'link\', \'link\' is a binary variable, link=1 if there\'