def Compute_precision_recall_gain(self):
'''One might choose PRG if there is little interest in identifying false negatives '''
prg_curve = prg.create_prg_curve(self.y_test, self.y_score)
auprg = prg.calc_auprg(prg_curve)
prg.plot_prg(prg_curve)
logging.info('auprg: %.3f' % auprg)
logging.info("compute_precision_recall_gain Complete")
pass
def Compute_precision_recall_gain(self):
'''One might choose PRG if there is little interest in identifying false negatives '''
for count,sim in enumerate(self.sim_list):
prg_curve = prg.create_prg_curve(self.y_test, self.y_score[count])
auprg = prg.calc_auprg(prg_curve)
prg.plot_prg(prg_curve)
logging.info('auprg: %.3f' % auprg)
logging.info("compute_precision_recall_gain Complete: "+str(sim))
pass
def do_prg(scores, true_labels, file_name='', directory='', plot=False):
prg_curve = prg.create_prg_curve(true_labels, scores)
auprg = prg.calc_auprg(prg_curve)
if plot:
prg.plot_prg(prg_curve)
plt.title('Precision-Recall-Gain curve: AUC=%0.4f' % (auprg))
plt.savefig(directory + file_name + "prg.png")
return auprg