pre_rec_equal.m

来自「一个包含丰富内容的流形学习算法工具包」· M 代码 · 共 13 行

function obj=pre_rec_equal(np,nn,fp,fn);

% use with breakeven.m
% this function maximizes when precision is close to recall
recall = (np-fn)/np;
  if fp+np-fn > 0
    precision = (np-fn)/(fp+np-fn);
  else
    precision = 0;
  end
 
     obj=1-(precision-recall)^2;