二分类: 逻辑回归 Fisher线性判别分析
多分类:spss 多分类线性判别分析 多分类逻辑回归
二分类
逻辑回归
![](https://www.icode9.com/i/ll/?i=img_convert/04b8a83d89e8b1cf57703ad279c90212.png)
![](https://www.icode9.com/i/ll/?i=img_convert/ae7c7c09f9fb982a5369e44d80c74b06.png)
![](https://www.icode9.com/i/ll/?i=img_convert/91fdadc775667cc0e949885d0f7ca256.png)
![](https://www.icode9.com/i/ll/?i=img_convert/bbb97191fc6ae01a1d87c847f2e84cd1.png)
![](https://www.icode9.com/i/ll/?i=img_convert/a41eb6e1f8baf80f8fdfb956b8fac5cc.png)
![](https://www.icode9.com/i/ll/?i=img_convert/b417bafab6ef06a95288020ece9eab70.png)
虚拟变量
![](https://www.icode9.com/i/ll/?i=img_convert/5cbe568e416916589e3806337036e38d.png)
逻辑回归求解
![](https://www.icode9.com/i/ll/?i=img_convert/719ac9535d33329ddeb187b0d9d68463.png)
![](https://www.icode9.com/i/ll/?i=img_convert/2f658212a9c9f1a1be7b7f320683ce6d.png)
![](https://www.icode9.com/i/ll/?i=img_convert/845c627537f1c177684766586b99c3bc.png)
![](https://www.icode9.com/i/ll/?i=img_convert/cae52a74ed89ad219a5dd38f679ccb28.png)
分类变量
![](https://www.icode9.com/i/ll/?i=img_convert/899562dd3c134ae00d472880e26f0d57.png)
![](https://www.icode9.com/i/ll/?i=img_convert/fdad1d854a7f65cea9780c841d4708d6.png)
预测结果差?
![](https://www.icode9.com/i/ll/?i=img_convert/516d52ab991ca402fcff0dd364c2f434.png)
但是加入平方项后 显出性可能都不显著,存在过拟合现象
![](https://www.icode9.com/i/ll/?i=img_convert/3a7441a907eebb2945f9a4f77c37fce6.png)
如何确定合适的模型
![](https://www.icode9.com/i/ll/?i=img_convert/6245ab953cddd820c35364a66da4de1d.png)
Fisher线性判别分析
![](https://www.icode9.com/i/ll/?i=img_convert/ae30b88405c50a3463bf724617c96e0c.png)
核心问题:找到线性系数向量w
spss操作
![](https://www.icode9.com/i/ll/?i=img_convert/f421e001fe26cb6a93ccd2a242c69ea3.png)
![](https://www.icode9.com/i/ll/?i=img_convert/de5464c037b0d5431b1bddb71f6bf796.png)
多分类
![](https://www.icode9.com/i/ll/?i=img_convert/5b6cdb7171a2bab8bfb446968c61b0e2.png)
多元逻辑
标签:逻辑,--,模型,分类,建模,判别分析,线性,Fisher,回归
来源: https://blog.csdn.net/bossDDYY/article/details/122759392