基于数据驱动建模的毒情预测方法研究

摘要

Date drive modeling is a kind modeling method of not causal relationship, it aims at the fact that the traditional regressive model relies on the experience, involving the problems of subjective and blind factors, and it doesn＇ t need a theoretic model supported by theory （ or hypothesis） in the according to the analysis of the data structure, in the light of the change rule of data itself, trapolating mechanism to describe the change of time se] tity is less, the forecasting model of the fitting accuracy eral applicable modeling thoughts. modeling set stage, in view of using ex-%数据驱动建模是一种非因果关系建模方法，针对传统回归模型依赖于经验，含有主观和盲目因素的问题，在模型设定阶段不需要一个以既有理论（或假说）为支持的理论模型，通过对数据结构的分析，依据数据自身的变化规律，利用外推机制描述时间序列数据的变化，从而有效解决数据量较少、预测模型拟合精度较低的问题，由此提供了一条普遍适用的建模思路。