更新了部分文档

Day36-40/code/HRS_create_and_init.sql

@@ -25,14 +25,13 @@ job varchar(20) not null comment '员工职位',
 mgr int comment '主管编号',
 sal int not null comment '员工月薪',
 comm int comment '每月补贴',
-dno int comment '所在部门编号',
-primary key (eno),
-foreign key (dno) references tb_dept(dno),
-foreign key (mgr) references tb_emp(eno)
+dno int comment '所在部门编号'
 );
 
+alter table tb_emp add constraint pk_emp_eno primary key (eno);
+alter table tb_emp add constraint uk_emp_ename unique (ename);
 -- alter table tb_emp add constraint fk_emp_mgr foreign key (mgr) references tb_emp (eno);
--- alter table tb_emp add constraint fk_emp_dno foreign key (dno) references tb_dept (dno);
+alter table tb_emp add constraint fk_emp_dno foreign key (dno) references tb_dept (dno);
 
 insert into tb_emp values 
     (7800, '张三丰', '总裁', null, 9000, 1200, 20),
@@ -61,11 +60,11 @@ insert into tb_emp values
 
 -- 查询月薪最高的员工(Boss除外)的姓名和月薪
 
--- 查询薪水超过平均薪水的员工的姓名和月薪
+-- 查询月薪超过平均月薪的员工的姓名和月薪
 
--- 查询薪水超过其所在部门平均薪水的员工的姓名、部门编号和月薪
+-- 查询月薪超过其所在部门平均月薪的员工的姓名、部门编号和月薪
 
--- 查询部门中薪水最高的人姓名、月薪和所在部门名称
+-- 查询部门中月薪最高的人姓名、月薪和所在部门名称
 
 -- 查询主管的姓名和职位
 

Day36-40/code/SRS_create_and_init.sql

@@ -14,7 +14,7 @@ create table `tb_college`
 `col_name` varchar(50) not null comment '名称',
 `col_intro` varchar(5000) default '' comment '介绍',
 primary key (`col_id`)
-) engine=innodb;
+) engine=innodb comment '学院表';
 
 -- 创建学生表
 create table `tb_student`
@@ -27,7 +27,7 @@ create table `tb_student`
 `col_id` int unsigned not null comment '所属学院',
 primary key (`stu_id`),
 foreign key (`col_id`) references `tb_college` (`col_id`)
-) engine=innodb;
+) engine=innodb comment '学生表';
 
 -- 创建教师表
 create table `tb_teacher`
@@ -38,7 +38,7 @@ create table `tb_teacher`
 `col_id` int unsigned not null comment '所属学院',
 primary key (`tea_id`),
 foreign key (`col_id`) references `tb_college` (`col_id`)
-);
+) engine=innodb comment '老师表';
 
 -- 创建课程表
 create table `tb_course`
@@ -49,7 +49,7 @@ create table `tb_course`
 `tea_id` int unsigned not null comment '授课老师',
 primary key (`cou_id`),
 foreign key (`tea_id`) references `tb_teacher` (`tea_id`)
-);
+) engine=innodb comment '课程表';
 
 -- 创建选课记录表
 create table `tb_record`
@@ -63,7 +63,7 @@ primary key (`rec_id`),
 foreign key (`sid`) references `tb_student` (`stu_id`),
 foreign key (`cid`) references `tb_course` (`cou_id`),
 unique (`sid`, `cid`)
-);
+) engine=innodb comment '选课记录表';
 
 -- 插入学院数据
 insert into `tb_college` 

Day36-40/code/hrs_with_answer.sql

@@ -0,0 +1,144 @@
+drop database if exists hrs;
+create database hrs default charset utf8mb4;
+
+use hrs;
+
+create table tb_dept
+(
+dno int not null comment '编号',
+dname varchar(10) not null comment '名称',
+dloc varchar(20) not null comment '所在地',
+primary key (dno)
+);
+
+insert into tb_dept values 
+    (10, '会计部', '北京'),
+    (20, '研发部', '成都'),
+    (30, '销售部', '重庆'),
+    (40, '运维部', '深圳');
+
+create table tb_emp
+(
+eno int not null comment '员工编号',
+ename varchar(20) not null comment '员工姓名',
+job varchar(20) not null comment '员工职位',
+mgr int comment '主管编号',
+sal int not null comment '员工月薪',
+comm int comment '每月补贴',
+dno int comment '所在部门编号',
+primary key (eno),
+foreign key (dno) references tb_dept (dno)
+);
+
+-- alter table tb_emp add constraint pk_emp_eno primary key (eno);
+-- alter table tb_emp add constraint uk_emp_ename unique (ename);
+-- alter table tb_emp add constraint fk_emp_mgr foreign key (mgr) references tb_emp (eno);
+-- alter table tb_emp add constraint fk_emp_dno foreign key (dno) references tb_dept (dno);
+
+insert into tb_emp values 
+    (7800, '张三丰', '总裁', null, 9000, 1200, 20),
+    (2056, '乔峰', '分析师', 7800, 5000, 1500, 20),
+    (3088, '李莫愁', '设计师', 2056, 3500, 800, 20),
+    (3211, '张无忌', '程序员', 2056, 3200, null, 20),
+    (3233, '丘处机', '程序员', 2056, 3400, null, 20),
+    (3251, '张翠山', '程序员', 2056, 4000, null, 20),
+    (5566, '宋远桥', '会计师', 7800, 4000, 1000, 10),
+    (5234, '郭靖', '出纳', 5566, 2000, null, 10),
+    (3344, '黄蓉', '销售主管', 7800, 3000, 800, 30),
+    (1359, '胡一刀', '销售员', 3344, 1800, 200, 30),
+    (4466, '苗人凤', '销售员', 3344, 2500, null, 30),
+    (3244, '欧阳锋', '程序员', 3088, 3200, null, 20),
+    (3577, '杨过', '会计', 5566, 2200, null, 10),
+    (3588, '朱九真', '会计', 5566, 2500, null, 10);
+
+
+-- 查询月薪最高的员工姓名和月薪
+select ename, sal from tb_emp where sal=(select max(sal) from tb_emp);
+
+select ename, sal from tb_emp where sal>=all(select sal from tb_emp);
+
+-- 查询员工的姓名和年薪((月薪+补贴)*13)
+select ename, (sal+ifnull(comm,0))*13 as ann_sal from tb_emp order by ann_sal desc;
+
+-- 查询有员工的部门的编号和人数
+select dno, count(*) as total from tb_emp group by dno;
+
+-- 查询所有部门的名称和人数
+select dname, ifnull(total,0) as total from tb_dept left join 
+(select dno, count(*) as total from tb_emp group by dno) tb_temp 
+on tb_dept.dno=tb_temp.dno;
+
+-- 查询月薪最高的员工(Boss除外)的姓名和月薪
+select ename, sal from tb_emp where sal=(
+	select max(sal) from tb_emp where mgr is not null
+);
+
+-- 查询月薪排第2名的员工的姓名和月薪
+select ename, sal from tb_emp where sal=(
+	select distinct sal from tb_emp order by sal desc limit 1,1
+);
+
+select ename, sal from tb_emp where sal=(
+	select max(sal) from tb_emp where sal<(select max(sal) from tb_emp)
+);
+
+-- 查询月薪超过平均月薪的员工的姓名和月薪
+select ename, sal from tb_emp where sal>(select avg(sal) from tb_emp);
+
+-- 查询月薪超过其所在部门平均月薪的员工的姓名、部门编号和月薪
+select ename, t1.dno, sal from tb_emp t1 inner join 
+(select dno, avg(sal) as avg_sal from tb_emp group by dno) t2
+on t1.dno=t2.dno and sal>avg_sal;
+
+-- 查询部门中月薪最高的人姓名、月薪和所在部门名称
+select ename, sal, dname 
+from tb_emp t1, tb_dept t2, (
+	select dno, max(sal) as max_sal from tb_emp group by dno
+) t3 where t1.dno=t2.dno and t1.dno=t3.dno and sal=max_sal;
+
+-- 查询主管的姓名和职位
+-- 提示：尽量少用in/not in运算，尽量少用distinct操作
+-- 可以使用存在性判断（exists/not exists）替代集合运算和去重操作
+select ename, job from tb_emp where eno in (
+	select distinct mgr from tb_emp where mgr is not null
+);
+
+select ename, job from tb_emp where eno=any(
+	select distinct mgr from tb_emp where mgr is not null
+);
+
+select ename, job from tb_emp t1 where exists (
+	select 'x' from tb_emp t2 where t1.eno=t2.mgr
+);
+
+-- MySQL8有窗口函数：row_number() / rank() / dense_rank()
+-- 查询月薪排名4~6名的员工的排名、姓名和月薪
+select ename, sal from tb_emp order by sal desc limit 3,3;
+
+select row_num, ename, sal from 
+(select @a:=@a+1 as row_num, ename, sal 
+from tb_emp, (select @a:=0) t1 order by sal desc) t2 
+where row_num between 4 and 6;
+
+-- 窗口函数不适合业务数据库，只适合做离线数据分析
+select 
+	ename, sal, 
+	row_number() over (order by sal desc) as row_num,
+    rank() over (order by sal desc) as ranking,
+    dense_rank() over (order by sal desc) as dense_ranking
+from tb_emp limit 3 offset 3;
+
+select ename, sal, ranking from (
+	select ename, sal, dense_rank() over (order by sal desc) as ranking from tb_emp
+) tb_temp where ranking between 4 and 6;
+
+-- 窗口函数主要用于解决TopN查询问题
+-- 查询每个部门月薪排前2名的员工姓名、月薪和部门编号
+select ename, sal, dno from (
+	select ename, sal, dno, rank() over (partition by dno order by sal desc) as ranking
+	from tb_emp
+) tb_temp where ranking<=2;
+
+select ename, sal, dno from tb_emp t1 
+where (select count(*) from tb_emp t2 where t1.dno=t2.dno and t2.sal>t1.sal)<2 
+order by dno asc, sal desc;

Day66-70/66.数据分析概述.md

@@ -259,7 +259,7 @@ Notebook是基于网页的用于交互计算的应用程序，可以用于代码
 
             在实际工作中，我们经常通过四分位数再配合[箱线图](https://zhuanlan.zhihu.com/p/110580568)来发现异常值。例如，小于$Q_1 - 1.5 \times IQR$的值或大于$Q3 + 1.5 \times IQR$的值可以视为普通异常值，而小于$Q_1 - 3 \times IQR$的值或大于$Q3 + 3 \times IQR$的值通常视为极度异常值。这种检测异常值的方法跟[“3西格玛法则”](https://zh.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7%E5%8E%9F%E5%89%87)的道理是一致的，如下图所示。
 
-            ![](res/quartile_and_3sigma.png)
+            ![](res/3sigma.png)
 
 2. 离散趋势