- 下载安装必要的库
pip install pypmml
pip install sklearn2pmml
- 导入必要的库
from sklearn2pmml import PMMLPipeline, sklearn2pmml
from pypmml import Model
- 保存模型,以lr模型为例
pipeline = PMMLPipeline([
('classifier', LogisticRegression(C=gscv.best_params_['C'], class_weight=gscv.best_params_['class_weight'], solver='saga'))
])
pipeline.fit(x_train, y_train)
sklearn2pmml(pipeline, 'lr.pmml')
- 加载模型进行预测,打印指标
def ProbabilitytoValue(x):
if x >= 0.5:
return 1
return 0
modelpmml = Model.fromFile('lr.pmml')
ypre_pmml = modelpmml.predict(x_test)
ypre_pmml = list(map(ProbabilitytoValue, np.array(ypre_pmml)[:, 1]))
print(recall_score(y_test, ypre_pmml))
print(precision_score(y_test, ypre_pmml))
print(accuracy_score(y_test, ypre_pmml))