【hadoop】疫情离线分析案例

准备：

疫情数据集，mysql，hive，sqoop，davinci，hadoop集群环境

阶段1：数据准备​​

​​1. MySQL源表

mysql

CREATE DATABASE covid19 DEFAULT CHARSET 'utf8mb4'

CREATE TABLE `covid_source` (
  `id` INT NOT NULL AUTO_INCREMENT,
  `province` VARCHAR(20),
  `date` DATE,
  `confirmed` INT,
  `deaths` INT,
  `recovered` INT,
  PRIMARY KEY (`id`)
);

-- 插入数据语句（可以使用Python，也可以sql）
INSERT INTO `covid_source` 
(province, date, confirmed, deaths, recovered) VALUES
('湖北','2023-01-01', 54306, 1320, 32415),
('广东','2023-01-01', 1523, 8, 1240),
...

（你也可以使用LOAD DATA命令，直接将数据集导入到Hive中） 

阶段2：Hive数仓构建​​

​​1. Sqoop导入Hive（关键命令）​​

bash

sqoop import \
--connect jdbc:mysql://master:3306/covid19 \
--username root \
--password 123456 \
--table covid_source \
--hive-import \
--hive-table ods_covid_source \
--create-hive-table \
--fields-terminated-by '\001'

2. Hive分层建模​​

hivesql

-- ODS层（原始数据）（Sqoop导入时自动创建）
CREATE TABLE ods_covid_source (
  province STRING,
  `date` DATE,
  confirmed INT,
  deaths INT,
  recovered INT
) STORED AS TEXTFILE;

-- DWD层（清洗后）
CREATE TABLE dwd_covid_stats AS
SELECT 
  province,
  `date`,
  confirmed,
  deaths,
  recovered,
  (confirmed - deaths - recovered) AS current_confirmed 
FROM ods_covid_source 
WHERE `date` IS NOT NULL;

阶段3：分析指标计算​​

​​1. 核心分析SQL示例​​

hivesql

-- 各省累计确诊--DWS层
CREATE TABLE dws_province_top AS
SELECT 
  province,
  MAX(confirmed) AS total_confirmed				
FROM dwd_covid_stats
GROUP BY province
ORDER BY total_confirmed DESC;

-- 全国每日新增趋势--DWS层
CREATE TABLE dws_daily_growth AS
SELECT 
  `date`,
CASE
WHEN `date` = MIN(`date`) OVER() THEN 0
ELSE SUM(confirmed) - LAG(SUM(confirmed),1) OVER(ORDER BY `date`) 
END AS new_confirmed
FROM dwd_covid_stats
GROUP BY `date`;

阶段4：Sqoop导出到MySQL​​

​​1. MySQL结果表设计​​

mysql

CREATE TABLE `ads_province_top` (
  `province` VARCHAR(20) PRIMARY KEY,
  `total_confirmed` INT
);

CREATE TABLE `ads_daily_growth` (
  `date` DATE PRIMARY KEY,
  `new_confirmed` INT
);

2. Sqoop导出命令​​

bash

# 导出省份TOP10数据
sqoop export \
--connect "jdbc:mysql://master:3306/covid19?useUnicode=true&characterEncoding=utf8" \
--username root \
--password 123456 \
--table ads_province_top \
--export-dir /user/hive/warehouse/dws_province_top \
--input-fields-terminated-by '\001'

# 导出每日趋势数据（增量导出示例）
sqoop export \
--connect "jdbc:mysql://master:3306/covid19?useUnicode=true&characterEncoding=utf8" \
--username root \
--password 123456 \
--table ads_daily_growth \
--export-dir /user/hive/warehouse/dws_daily_growth \
--input-fields-terminated-by '\001'

阶段5：Davinci可视化配置​​

​​1. 数据源连接​​

• 在Davinci中添加MySQL数据源：

类型: MySQL
URL: jdbc:mysql://master:3306/covid19
用户名/密码: root 123456

2. 可视化看板设计​​

• ​​组件1：疫情地图​​
  • 数据模型：SELECT province, total_confirmed FROM ads_province_top
  • 图表类型：中国地图
  • 映射字段：province -> 区域, total_confirmed -> 颜色深浅
• ​​组件2：趋势折线图​​
  • 数据模型：SELECT date, new_confirmed FROM ads_daily_growth
  • 图表类型：折线图
  • X轴：date, Y轴：new_confirmed

效果图