第12名方案后处理

raters = np.array([18, 18,  6,  6,  6,  6, 18, 18,  6,  6,  6,  6,  6,  6,  6,  6,  6, 6,  6,  3, 18, 18, 18, 18, 18, 90,  6,  6,  6, 18])

mins = np.min(all_preds,axis=0)
maxs = np.max(all_preds,axis=0)
all_preds = (all_preds - mins)/(maxs - mins)

all_preds =  np.round(raters*all_preds).astype(np.float)/raters

如果你有时间，请尝试将此方法应用到你的提交结果中，我非常有兴趣知道这能为你的分数带来多少提升。