复杂数据下基于随机森林和决策树的辛烷值损失预测模型

Octane Loss Prediction Model Based on Random Forest and Decision Tree under Complex Data

燃料清洁化要求对汽油进行脱硫和降烯烃处理,同时尽量保持其辛烷值。由于现有数据具有缺失、高维等特征,笔者提出一种基于随机森林和决策树的辛烷值损失预测框架,采用缺失比率分析、数据空缺填补、方差过滤和异常值处理4个方法对数据进行预处理,并利用随机森林筛选25个变量作为主要变量。

Fuel cleaning requires desulfurization and olefin reduction of gasoline, while maintaining its octane number as much as possible. Because the existing data has the characteristics of missing and high dimension, the author proposes an octane loss prediction framework based on random forest and decision tree. The data are preprocessed by four methods: missing ratio analysis,data vacancy filling, variance filtering and outlier processing, and 25 variables are selected by random forest as the main variables.