gentang / intro_ds Goto Github PK

def evaluation(re):
    """
    计算预测结果的查准查全率以及f1
    参数
    ----
    re ：DataFrame，预测结果，里面包含两列：真实值‘lable_code’、预测值‘pred’
    """
    bins = np.array([0, 0.5, 1])
    label = re["label_code"]
    pred = re["pred"]
    tp, fp, fn, tn = np.histogram2d(label, pred, bins=bins)[0].flatten()
    precision = tp / (tp + fp)  # 0.951
    recall = tp / (tp + fn)  # 0.826
    f1 = 2 * precision * recall / (precision + recall)  # 0.884
    print("查准率: %.3f, 查全率: %.3f, f1: %.3f" % (precision, recall, f1))

bins=np.array([0, 0.5, 1])
x=pd.Series([0, 1, 0, 1, 1, 1, 0])
y=pd.Series([1, 1, 0, 0, 1, 1, 1])
a, b, c, d = np.histogram2d(x,y,bins=bins)[0].flatten()

In [119]: a,b,c,d
Out[119]: (1.0, 2.0, 1.0, 3.0)