hive sql 语句学习

use liuliangyuan680_ind;

--drop table llid;

create table llid(insurance_no string,group_no int) ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t' STORED AS TEXTFILE; 

load data  inpath '/apps-data/tmp/gbd-ide/hive-liuliangyuan680/file/id.txt'

into table liuliangyuan680_ind.llid;

select * from liuliangyuan680_ind.llid;

select count(insurance_no) from liuliangyuan680_ind.llid;

SELECT * from liuliangyuan680_ind.llid a , liuliangyuan680_ind.rxw_xmsb_djmxypb b where a.insurance_no =b.insurance_no;

select * from liuliangyuan680_ind.rxw_xmsb_djmxypb;

use liuliangyuan680_ind;

--drop table summary3month;

CREATE table summary3month AS

select insurance_no, sum( case when visit_type='1' and out_hosp_date is not null and in_hosp_date is not null then  (datediff(out_hosp_date,in_hosp_date)+1) else 0 end) hosp_days, 

--总频次

COUNT(DISTINCT(case when visit_type='1' then visit_no else null end)) hosp_freq,

COUNT(DISTINCT(case when visit_type='2' then visit_no else null end)) visit_freq,

COUNT(DISTINCT(case when diagnose_code is not null then diagnose_code else null end)) diagnose_num,COUNT(distinct medical_org_code) as org_num,count(DISTINCT bill_no) as bill_num,count(DISTINCT bill_detail_no) as detail_num,

--看医生数,看病月数

count (distinct doctor_code) as doctor_num,

count (distinct month(in_hosp_date)) as month_num,

--住院药品数，诊疗项目数，服务设施数,总金额，总赔付

count (distinct (case when (visit_type="1" and project_type="1") then project_name else null end)) as hosp_medicine_num,

count (distinct (case when (visit_type="1" and project_type="2") then project_name else null end)) as hosp_check_num,

count (distinct (case when (visit_type="1" and project_type="3") then project_name else null end)) as hosp_service_num,

sum(case when visit_type='1' then mx_sum_amount else null end) as hosp_sum_amount,

sum(case when visit_type='1' then mx_apply_pay_amount else null end) as hosp_apply_amount,

--门诊药品数，诊疗项目数，服务设施数,总金额，总赔付

count (distinct (case when (visit_type="2" and project_type="1") then project_name else null end)) as visit_medicine_num,

count (distinct (case when (visit_type="2" and project_type="1") then bill_detail_no else null end)) as visit_medicine_freq,

count (distinct (case when (visit_type="2" and project_type="2") then project_name else null end)) as visit_check_num,

count (distinct (case when (visit_type="2" and project_type="2") then bill_detail_no else null end)) as visit_check_freq,

count (distinct (case when (visit_type="2" and project_type="3") then project_name else null end)) as visit_service_num,

count (distinct (case when (visit_type="2" and project_type="3") then bill_detail_no else null end)) as v
4000
isit_service_freq,

sum(case when visit_type='2' then mx_sum_amount else null end) as visit_sum_amount,

sum(case when visit_type='2' then mx_apply_pay_amount else null end) as visit_apply_amount

from liuliangyuan680_ind.rxw_xmsb_djmxypb 

where (visit_type = '1' and 

       year(in_hosp_date) ="2014" ) or 

       (visit_type = '2' and year(in_hosp_date) ="2014" ) 

group BY insurance_no;

use liuliangyuan680_ind;

--drop table summary_full_201410;

CREATE table summary_full_201410 AS

select insurance_no, sum( case when visit_type='1' and out_hosp_date is not null and in_hosp_date is not null then  (datediff(out_hosp_date,in_hosp_date)+1) else 0 end) hosp_days,

COUNT(DISTINCT(case when visit_type='1' then visit_no else null end)) hosp_freq,

COUNT(DISTINCT(case when visit_type='2' then visit_no else null end)) visit_freq,

COUNT(DISTINCT(case when diagnose_code is not null then diagnose_code else null end)) diagnose_num,COUNT(distinct medical_org_code) as org_num,count(DISTINCT bill_no) as bill_num,count(DISTINCT bill_detail_no) as detail_num,

--看医生数

count (distinct doctor_code) as doctor_num,

--住院药品数，诊疗项目数，服务设施数,总金额，总赔付

count (distinct (case when (visit_type="1" and project_type="1") then project_name else null end)) as hosp_medicine_num,

count (distinct (case when (visit_type="1" and project_type="2") then project_name else null end)) as hosp_check_num,

count (distinct (case when (visit_type="1" and project_type="3") then project_name else null end)) as hosp_service_num,

sum(case when visit_type='1' then mx_sum_amount else null end) as hosp_sum_amount,

sum(case when visit_type='1' then mx_apply_pay_amount else null end) as hosp_apply_amount,

--门诊药品数，诊疗项目数，服务设施数,总金额，总赔付

count (distinct (case when (visit_type="2" and project_type="1") then project_name else null end)) as visit_medicine_num,

count (distinct case when (visit_type="2" and project_type="1") then bill_detail_no else null end)) as visit_medicine_freq,

count (distinct (case when (visit_type="2" and project_type="2") then project_name else null end)) as visit_check_num,

count (distinct (case when (visit_type="2" and project_type="2") then bill_detail_no else null end)) as visit_check_freq,

count (distinct (case when (visit_type="2" and project_type="3") then project_name else null end)) as visit_service_num,

count (distinct (case when (visit_type="2" and project_type="3") then bill_detail_no else null end)) as visit_service_freq,

sum(case when visit_type='2' then mx_sum_amount else null end) as visit_sum_amount,

sum(case when visit_type='2' then mx_apply_pay_amount else null end) as visit_apply_amount

from liuliangyuan680_ind.rxw_xmsb_djmxypb 

where (visit_type = '1' and 

        year(in_hosp_date) ="2014" and 

        month(in_hosp_date)="11") or 

        (visit_type = '2' and year(in_hosp_date) ="2014" and 

        month(in_hosp_date)="11") 

group BY insurance_no;

use liuliangyuan680_ind;

--drop table sefinal;

create table sefinal as SELECT

b.insurance_no,

b.medical_org_name,

b.medical_dept_name,

b.doctor_code

from liuliangyuan680_ind.llid a , liuliangyuan680_ind.rxw_xmsb_djmxypb b where a.insurance_no =b.insurance_no;

select * from liuliangyuan680_ind.sefinal;

insert overwrite local directory '/home/wyp/wyp' select * from liuliangyuan680_ind.selfinal; 

use liuliangyuan680_ind;

drop table llid;

create table llid(insurance_no string, group_no int) ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t' STORED AS TEXTFILE; 

load data  inpath '/apps-data/tmp/gbd-ide/hive-liuliangyuan680/file/id.txt'

into table liuliangyuan680_ind.llid;

select * from liuliangyuan680_ind.llid;

select count(insurance_no) from liuliangyuan680_ind.llid;

use liuliangyuan680_ind;

--drop table sefinal;

create table sefinal as SELECT

a.group_no,

b.insurance_no,

b.diagnose_code,

b.project_name,

b.medical_org_name,

b.medical_dept_name,

b.doctor_code

from liuliangyuan680_ind.llid a , liuliangyuan680_ind.rxw_xmsb_djmxypb b where a.insurance_no =b.insurance_no;

use liuliangyuan680_ind;

--drop table sefinal;

create table selfinal as SELECT

a.group_no,

b.insurance_no,

b.diagnose_code,

b.project_name,

b.medical_org_name,

b.medical_dept_name,

b.doctor_code

from liuliangyuan680_ind.llid a join liuliangyuan680_ind.rxw_xmsb_djmxypb b on ( a.insurance_no =b.insurance_no);

select * from liuliangyuan680_ind.sefinal;

select count(DISTINCT insurance_no) from liuliangyuan680_ind.sefinal;

---社保号个数1484,群数10

select count(DISTINCT insurance_no) from liuliangyuan680_ind.selfinal;

select count(DISTINCT group_no) from liuliangyuan680_ind.selfinal;

select  medical_org_name,COUNT(medical_org_name) as time1 FROM liuliangyuan680_ind.selfinal a WHERE

a.group_no=1 group by medical_org_name order by time1 ;

select  medical_org_name,COUNT(medical_org_name) FROM liuliangyuan680_ind.selfinal  

group by group_no and medical_org_name;

select  medical_org_name, COUNT(medical_org_name) FROM liuliangyuan680_ind.selfinal

where group_no=1;