scikit-learn的机器学习流水线技术与应用

Principle and Application of Scikit-Learn’s Machine Learning Pipeline

为运用流水线技术组合机器学习的步骤,并行运行程序,选择更优的学习模型,文章剖析了scikit-learn提供的机器学习流水线技术原理。运用scikit-learn的机器学习流水线技术构建了多阶多项式的流水线,应用该流水线绘制了1~6阶多项式的学习曲线,经对比分析,选择了拟合出的最优的3阶多项式,其收敛值为0.962,曲线图形与源数据曲线图形较接近。

To use pipeline technology to combine machine learning steps,run programs in parallel,and select a better learning model,this paper analyzes the principle of machine learning pipeline provided by scikit-learn.By using the machine learning pipeline of scikit-learn,a multi-order polynomial pipeline is constructed.The learning curve of 1-6 order polynomial is drawn by using the pipeline.After comparative analysis,the optimal third-order polynomial is selected,the convergence value is 0.966,and the curve graph is close to the source data curve graph.