基于交互式逐步回归的双语教学态度预测模型

针对多元线性回归分析法预测双语教学态度误差较大的问题,通过引入变量的平方项及交叉乘积项建立拟线性回归方程,利用交互式逐步回归分析法对变量进行筛选,并以筛选后的变量建模。将调查数据分为建模样本数据和测试样本数据,测试结果表明经过变量扩维及筛选所建立的预测模型精确度有较大提高,为语言学研究提供了一种新方法,值得推扩和借鉴。

Interactive stepwise regression prediction model based on bilingual teaching attitude

On the question that multiple linear regression analysis to predict bilingual teaching attitude causes big error, this paper uses interactive stepwise regression analysis to filter the variables and then modeling by introducing a variable quadratic component and cross-product terms to establish quasi-linear regression equation. The survey data is divided into modeling sample data and test sample data. The test results show that the accuracy of established prediction model through augmented and selected variable has improved greatly.This provides a new method for linguistic research; it is worth expanding and reference.