from sklearn import metrics
y_true = [0, 1, 2, 0, 1, 2,1]
y_pred = [0, 2, 1, 0, 0, 1,1]
a = metrics.precision_score(y_true, y_pred, average='weighted')
b = metrics.recall_score(y_true, y_pred, average='weighted')
c = metrics.f1_score(y_true, y_pred, average='weighted')
print(a,b,c,2*a*b/(a+b))
micro 为不分类别进行计算,
macro 分母为类别总量,例子中为3
sample 分母为样本总量,例子中为7
weight 按照各类别的样本量为权重进行计算
http://scikit-learn.org/stable/modules/model_evaluation.html#multiclass-and-multilabel-classification