机器学习 2 （KNN实例：预测年收入、保存模型）

小灰灰 2023-10-08 12:04 44阅读 0赞

#### KNN实例：预测年收入 ####

*   *   *  1、预测年收入
         *   *  1.1、导入需要的包
             *  1.2、准备数据y和目标值X
             *  1.3、转换str类型的数据
             *  1.4、分割数据
             *  1.5、训练数据并测试
             *  1.6、数据归一化
             *   *  1.6.1、数据归一化手动写代码
                 *  1.6.2、数据归一化包
         *  2、保存模型

#### 1、预测年收入 ####

首先我们准备一个带有年收入信息的表  
![在这里插入图片描述][2020030312452199.png]  
这个表里面自带每年收入值

##### 1.1、导入需要的包 #####

![在这里插入图片描述][20200303110047387.png]  
读取文件  
![在这里插入图片描述][20200303110135552.png]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70]

##### 1.2、准备数据y和目标值X #####

y：我们选取salary（这个salary是前面的取变量名）的最后一列数据salary（这个salary是列名）

X：目标值  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 1]  
查看X每列的数据类型 ：我们发现里面含有str类型的数据，是不能计算距离的  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 2]

##### 1.3、转换str类型的数据 #####

先查看X【“workclass”】里面有哪些数据  
![在这里插入图片描述][20200303122000542.png]  
先给不同的值组合一个索引  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 3]  
数据转换：根据提供的索引做映射  
![在这里插入图片描述][20200303124218350.png]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 4]  
通过查看表我们发现从第三列开始到倒数第二列都需要转换数据  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 5]  
接下来将这些需要修改数据类型的列进行批量修改

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 6]  
查看修改后的X  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 7]

##### 1.4、分割数据 #####

导包  
![在这里插入图片描述][20200303134311925.png]  
切割X、y  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 8]

##### 1.5、训练数据并测试 #####

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 9]  
提高邻居数量试试  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 10]

##### 1.6、数据归一化 #####

###### 1.6.1、数据归一化手动写代码 ######

数据归一化：消除差异性  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 11]

> 数据归一化：https://blog.csdn.net/haoji007/article/details/81157224

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 12]  
将归一化的数据进行训练和预测：准确率明显提高  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 13]  
数据归一化第二种方法  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 14]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 15]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 16]

###### 1.6.2、数据归一化包 ######

导包  
![在这里插入图片描述][20200303145147452.png]  
Z-score归一化  
![在这里插入图片描述][20200303145243802.png]

min-max归一化![在这里插入图片描述][20200303145325410.png]

#### 2、保存模型 ####

![在这里插入图片描述][20200303152037694.png]  
保存  
![在这里插入图片描述][20200303151104449.png]  
加载看看  
![在这里插入图片描述][20200303151204648.png]  
测试数据准确率  
![在这里插入图片描述][20200303151215813.png]  
参数comress压缩级别  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 17]

> 本文学习笔记截至P515，18:12

[2020030312452199.png]: https://img-blog.csdnimg.cn/2020030312452199.png
[20200303110047387.png]: https://img-blog.csdnimg.cn/20200303110047387.png
[20200303110135552.png]: https://img-blog.csdnimg.cn/20200303110135552.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/2020030311141757.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20200303112128667.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20200303112232524.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[20200303122000542.png]: https://img-blog.csdnimg.cn/20200303122000542.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 3]: https://img-blog.csdnimg.cn/20200303123922723.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[20200303124218350.png]: https://img-blog.csdnimg.cn/20200303124218350.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 4]: https://img-blog.csdnimg.cn/20200303124237246.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 5]: https://img-blog.csdnimg.cn/2020030312545841.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 6]: https://img-blog.csdnimg.cn/20200303131620843.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 7]: https://img-blog.csdnimg.cn/20200303131656230.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[20200303134311925.png]: https://img-blog.csdnimg.cn/20200303134311925.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 8]: https://img-blog.csdnimg.cn/20200303134530617.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 9]: https://img-blog.csdnimg.cn/2020030313502664.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 10]: https://img-blog.csdnimg.cn/20200303140706774.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 11]: https://img-blog.csdnimg.cn/20200303142805552.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 12]: https://img-blog.csdnimg.cn/20200303143022133.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 13]: https://img-blog.csdnimg.cn/20200303143352523.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 14]: https://img-blog.csdnimg.cn/20200303144453669.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 15]: https://img-blog.csdnimg.cn/20200303144547227.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 16]: https://img-blog.csdnimg.cn/20200303144505803.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70
[20200303145147452.png]: https://img-blog.csdnimg.cn/20200303145147452.png
[20200303145243802.png]: https://img-blog.csdnimg.cn/20200303145243802.png
[20200303145325410.png]: https://img-blog.csdnimg.cn/20200303145325410.png
[20200303152037694.png]: https://img-blog.csdnimg.cn/20200303152037694.png
[20200303151104449.png]: https://img-blog.csdnimg.cn/20200303151104449.png
[20200303151204648.png]: https://img-blog.csdnimg.cn/20200303151204648.png
[20200303151215813.png]: https://img-blog.csdnimg.cn/20200303151215813.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f_size_16_color_FFFFFF_t_70 17]: https://img-blog.csdnimg.cn/20200303151806393.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2FfX2ludF9f,size_16,color_FFFFFF,t_70