• 企业400电话
  • 微网小程序
  • AI电话机器人
  • 电商代运营
  • 全 部 栏 目

    企业400电话 网络优化推广 AI电话机器人 呼叫中心 网站建设 商标✡知产 微网小程序 电商运营 彩铃•短信 增值拓展业务
    MYSQL每隔10分钟进行分组统计的实现方法

    前言

    本文的内容主要是介绍了MYSQL每隔10分钟进行分组统计的实现方法,在画用户登录、操作情况在一天内的分布图时会非常有用,之前我只知道用「存储过程」实现的方法(虽然执行速度快,但真的是太不灵活了),后来学会了用高级点的「group by」方法来灵活实现类似功能。

    正文:

    -- time_str '2016-11-20 04:31:11'
    -- date_str 20161120
    
    select concat(left(date_format(time_str, '%y-%m-%d %h:%i'),15),'0') as time_flag, count(*) as count from `security`.`cmd_info` where `date_str`=20161120 group by time_flag order by time_flag; -- 127 rows
    
    select round(unix_timestamp(time_str)/(10 * 60)) as timekey, count(*) from `security`.`cmd_info` where `date_str`=20161120 group by timekey order by timekey; -- 126 rows
    
    -- 以上2个SQL语句的思路类似——使用「group by」进行区分,但是方法有所不同,前者只能针对10分钟(或1小时)级别,后者可以动态调整间隔大小,两者效率差不多,可以根据实际情况选用
    
    select concat(date(time_str),' ',hour(time_str),':',round(minute(time_str)/10,0)*10), count(*) from `security`.`cmd_info` where `date_str`=20161120 group by date(time_str), hour(time_str), round(minute(time_str)/10,0)*10; -- 145 rows
    
    select concat(date(time_str),' ',hour(time_str),':',floor(minute(time_str)/10)*10), count(*) from `security`.`cmd_info` where `date_str`=20161120 group by date(time_str), hour(time_str), floor(minute(time_str)/10)*10; -- 127 rows (和 date_format 那个等价)
    
    select concat(date(time_str),' ',hour(time_str),':',ceil(minute(time_str)/10)*10), count(*) from `security`.`cmd_info` where `date_str`=20161120 group by date(time_str), hour(time_str), ceil(minute(time_str)/10)*10; -- 151 rows


    DELIMITER //
    
    DROP PROCEDURE IF EXISTS `usp_cmd_info`;
    
    CREATE PROCEDURE `usp_cmd_info`(IN dates VARCHAR(12))
    BEGIN
     SELECT count(*) from `cmd_info` where `time_str` BETWEEN CONCAT(dates, " 00:00:00") AND CONCAT(dates, " 00:10:00") INTO @count_0;
     SELECT count(*) from `cmd_info` where `time_str` BETWEEN CONCAT(dates, " 00:10:00") AND CONCAT(dates, " 00:20:00") INTO @count_1;
     ...
     SELECT count(*) from `cmd_info` where `time_str` BETWEEN CONCAT(dates, " 23:40:00") AND CONCAT(dates, " 23:50:00") INTO @count_142;
     SELECT count(*) from `cmd_info` where `time_str` BETWEEN CONCAT(dates, " 23:50:00") AND CONCAT(dates, " 23:59:59") INTO @count_143;
     select @count_0, @count_1, @count_2, @count_3, @count_4, @count_5, @count_6, @count_7, @count_8, @count_9, @count_10, @count_11, @count_12, @count_13, @count_14, @count_15, @count_16, @count_17, @count_18, @count_19, @count_20, @count_21, @count_22, @count_23, @count_24, @count_25, @count_26, @count_27, @count_28, @count_29, @count_30, @count_31, @count_32, @count_33, @count_34, @count_35, @count_36, @count_37, @count_38, @count_39, @count_40, @count_41, @count_42, @count_43, @count_44, @count_45, @count_46, @count_47, @count_48, @count_49, @count_50, @count_51, @count_52, @count_53, @count_54, @count_55, @count_56, @count_57, @count_58, @count_59, @count_60, @count_61, @count_62, @count_63, @count_64, @count_65, @count_66, @count_67, @count_68, @count_69, @count_70, @count_71, @count_72, @count_73, @count_74, @count_75, @count_76, @count_77, @count_78, @count_79, @count_80, @count_81, @count_82, @count_83, @count_84, @count_85, @count_86, @count_87, @count_88, @count_89, @count_90, @count_91, @count_92, @count_93, @count_94, @count_95, @count_96, @count_97, @count_98, @count_99, @count_100, @count_101, @count_102, @count_103, @count_104, @count_105, @count_106, @count_107, @count_108, @count_109, @count_110, @count_111, @count_112, @count_113, @count_114, @count_115, @count_116, @count_117, @count_118, @count_119, @count_120, @count_121, @count_122, @count_123, @count_124, @count_125, @count_126, @count_127, @count_128, @count_129, @count_130, @count_131, @count_132, @count_133, @count_134, @count_135, @count_136, @count_137, @count_138, @count_139, @count_140, @count_141, @count_142, @count_143;
    END //
    
    DELIMITER ;
    
    show PROCEDURE status\G
    
    CALL usp_cmd_info("2016-10-20");

    上面的这段MySQL存储过程的语句非常长,不可能用手工输入,可以用下面的这段Python代码按所需的时间间隔自动生成:

    import datetime
    
    today = datetime.date.today()
    # 或 由给定格式字符串转换成
    # today = datetime.datetime.strptime('2016-11-21', '%Y-%m-%d')
    
    min_today_time = datetime.datetime.combine(today, datetime.time.min) # 2016-11-21 00:00:00
    max_today_time = datetime.datetime.combine(today, datetime.time.max) # 2016-11-21 23:59:59
    
    sql_procedure_arr = []
    sql_procedure_arr2 = []
    for x in xrange(0, 60*24/5, 1):
      start_datetime = min_today_time + datetime.timedelta(minutes = 5*x)
      end_datetime = min_today_time + datetime.timedelta(minutes = 5*(x+1))
      # print x, start_datetime.strftime("%Y-%m-%d %H:%M:%S"), end_datetime.strftime("%Y-%m-%d %H:%M:%S")
      select_str = 'SELECT count(*) from `cmd_info` where `time_str` BETWEEN "{0}" AND "{1}" INTO @count_{2};'.format(start_datetime, end_datetime, x)
      # print select_str
      sql_procedure_arr.append(select_str)
      sql_procedure_arr2.append('@count_{0}'.format(x))
    print '\n'.join(sql_procedure_arr)
    print 'select {0};'.format(', '.join(sql_procedure_arr2))

    总结

    以上就是这篇文章的全部内容了,希望本文的内容对大家的学习或者工作能带来一定的帮助,如果有疑问大家可以留言交流。

    您可能感兴趣的文章:
    • MySQL 统计查询实现代码
    • MYSQL中有关SUM字段按条件统计使用IF函数(case)问题
    • 如何使用MySQL查询某个列中相同值的数量统计
    • 利用MySQL统计一列中不同值的数量方法示例
    • 单个select语句实现MySQL查询统计次数
    • MySQL中几种数据统计查询的基本使用教程
    • mysql按照天统计报表当天没有数据填0的实现代码
    • MySQL按时间统计数据的方法总结
    • mysql 获取规定时间段内的统计数据
    • MySQL统计函数GROUP_CONCAT使用陷阱分析
    • mysql滑动订单问题原理与解决方法实例分析
    上一篇:详解CentOS6.5 安装 MySQL5.1.71的方法
    下一篇:MySQL 四种事务隔离级别详解及对比
  • 相关文章
  • 

    © 2016-2020 巨人网络通讯 版权所有

    《增值电信业务经营许可证》 苏ICP备15040257号-8

    MYSQL每隔10分钟进行分组统计的实现方法 MYSQL,每隔,10分钟,进行,分组,