SQL query involving partial group with condition? - mysql

Below is a SQL query problem for which I am not able to understand correct approach:
DB tables:
Employee: emp_id, emp_name
Credit: credit_id, emp_id, credit_date, credit_amount
debit: debit_id, emp_id, debit_date, debit_amount
Here, each person can have multiple incomes and expenses.
Query requirement: At the end of each day, each employee will have some asset('credit till now' - 'debit till now'). We need to find top five employees in terms of maximum asset and the date on which they had this maximum asset.
I have tried the below query but seems like I am missing something:
select Credit.emp_id, Credit.date, (Credit.income_amount - Debit.credit_amount) from
(select emp_id, sum(amount) as credit_amount
from credit) Credit
LEFT JOIN LATERAL (
select emp_id, sum(amount) as debit_amount
from debits
where debits.emp_id = Credit.emp_id and Credit.date >= debits.date
group by debits.emp_id
) Debit
ON true

Here I'm breaking the query to make it more readable.
First of all, we need to get the total amount on a day-level for both credit and debit both, so that we can join the credit and debit table on the day level with the same emp_id.
with
credit as(
select emp_id,credit_date date,sum(credit_amount) as amount
from credit
group by 1,2),
debit as(
select emp_id,debit_date,sum(debit_amount) as amount
from expenses
group by 1,2),
Now we need to full outer join the "credit" and "debit" subqueries
payments as (
select distinct
case when c.emp_id is null then d.person_id else c.emp_id end as emp_id ,
case when c.emp_id is null then d.date else c.date end as date,
case when c.emp_id is null then 0 else i.amount end as credit ,
case when d.emp_id is null then 0 else d.amount end as debit
from credit c
full outer join debit d on d.emp_id=c.emp_id and d.date=c.date
),
Now we will take day-wise cumulative sum for credit, debit and total balance as shown below.
total_balance as(
SELECT emp_id, date,
sum(credit) OVER (PARTITION BY emp_id ORDER BY date asc) AS total_credit,
sum(debit) OVER (PARTITION BY emp_id ORDER BY date asc) AS total_debit,
(sum(income) OVER (PARTITION BY person_id ORDER BY date asc) -
sum(expense) OVER (PARTITION BY person_id ORDER BY date asc)) as total_balance
FROM group_payment
ORDER BY person_id, date),
Now we need to use the rank() function to assign rank based on total balance (desc) for an emp_id (ie. rank=1 will be assigned to the largest total balance on a day for a particular emp_id). The query is shown below.
ranks as (select emp_id,date,total_balance,
rank() over (partition by emp_id order by total_balance desc) as rank
from total_balance ),
Now pick the rows having rank=1 (ie. MAX of total_balance on a day for an emp_id and the date on which it was MAX).
Order it by total_balance descending and pick the top 5 rows
emp_order as (select emp_id,date,total_balance
from ranks
where rank=1
order by 3 desc
limit 5)
Now pick the name from the employee table.
select emp_id,name, date, total_balance as balance
from emp_order eo
join Employee e on e.emp_id = eo.emp_id
order by 4 desc

Group by and sum allows you to get the total credit for each person into 1 record. You can do a similar thing in a subquery to subtract the debit.
Select top 5 emp_id, credit_date, (sum(credit_amount) -
(select sum(debit_amount) from debit d
where c.emp_id = d.emp_id and c.credit_date = d.debit_date)
) as total
from Credit c group by emp_id, credit_date order by total

Related

how to retrieve the difference between the two given dates in SQL

How to find the Longest Booking ID for the given two dates and Costliest Booking ID for the given cost.
Here we have the 13 days difference so we are getting the longest booking id as 1.
what are the approach to archive this using sql query.
this will work:
(SELECT 'Total Booking Count' AS Label, COUNT(*) AS Value FROM bookings)
UNION ALL
(SELECT 'Longest Booking Id', booking_id FROM bookings ORDER BY DATEDIFF(enddate, startdate) DESC LIMIT 1)
UNION ALL
(SELECT 'Costliest Booking Id', booking_id FROM bookings ORDER BY (tariff*DATEDIFF(enddate, startdate)) DESC LIMIT 1)
with data
as (select *
,row_number() over(order by datediff(dd,end_date,start_date) desc) as rnk_time
,row_number() over(order by tarrif desc) as rnk_cost
,count(*) over(partition by 1) as tot_cnt
from your_table
)
select 'Total Booking count',tot_cnt
from data
where rnk_time=1
union all
select 'Longest Booking id',booking_id
from data
where rnk_time=1
union all
select 'Costliest Booking id',booking_id
from data
where rnk_cost=1

MySQL Select Top Row Grouping By Another Row

I have a sales table and I want to get each members most frequently shopped store in the last 3 months. The following query will get the every member with every store, but I want just one store per member.
SELECT member_id, store_id, COUNT(DISTINCT docket) as docket_count, SUM(dollar_amount) as dollars
FROM sales
WHERE TIMESTAMPDIFF(MONTH, sale_date, CURDATE()) < 3
GROUP BY member_id, store_id
ORDER BY member_id, docket_count DESC, dollars DESC
Or to get the top store for a single member
SELECT store_id, COUNT(DISTINCT docket) as docket_count, SUM(dollar_amount) as dollars
FROM sales
WHERE TIMESTAMPDIFF(MONTH, sale_date, CURDATE()) < 3
AND member_id = 1
GROUP BY store_id
ORDER BY docket_count DESC, dollars DESC
This is tricky. In MySQL, this can be easiest using the group_concat()/substring_index() trick:
SELECT member_id,
SUBSTRING_INDEX(GROUP_CONCAT(store_id ORDER BY docket_count DESC dollars DESC), ',', 1) as Most_Common_Store
FROM (SELECT member_id, store_id, COUNT(DISTINCT docket) as docket_count,
SUM(dollar_amount) as dollars
FROM sales
WHERE sale_date >= CURDATE() - interval 3 month
GROUP BY member_id, store_id
) ms
GROUP BY member_id;

How to Group a table and get results for a row based on the previous rows' data

I have a lookup table that relates dates and people associated with those dates:
id, user_id,date
1,1,2014-11-01
2,2,2014-11-01
3,1,2014-11-02
4,3,2014-11-02
5,1,2014-11-03
I can group these by date(day):
SELECT DATE_FORMAT(
MIN(date),
'%Y/%m/%d 00:00:00 GMT-0'
) AS date,
COUNT(*) as count
FROM user_x_date
GROUP BY ROUND(UNIX_TIMESTAMP(created_at) / 43200)
But, how can get the number of unique users, that have now shown up previously? For instance this would be a valid result:
unique, non-unique, date
2,0,2014-11-01
1,1,2014-11-02
0,1,2014-11-03
Is this possibly without having to rely on a scripting language to keep track of this data?
I think this query will do what you want, at least it seems to work for your limited sample data.
The idea is to use a correlated sub-query to check if the user_id has occurred on a date before the date of the current row and then do some basic arithmetic to determine number of unique/non-unique users for each date.
Please give it a try.
select
sum(u) - sum(n) as "unique",
sum(n) as "non-unique",
date
from (
select
date,
count(user_id) u,
case when exists (
select 1
from Table1 i
where i.user_id = o.user_id
and i.date < o.date
) then 1 else 0
end n
from Table1 o
group by date, user_id
) q
group by date
order by date;
Sample SQL Fiddle
I didn't include the id column in the sample fiddle as it's not needed (or used) to produce the result and won't change anything.
This is the relevant question: "But, how can get the number of unique users, that have now shown up previously?"
Calculate the first time a person shows up, and then use that for the aggregation:
SELECT date, count(*) as FirstVisit
FROM (SELECT user_id, MIN(date) as date
FROM user_x_date
GROUP BY user_id
) x
GROUP BY date;
I would then use this as a subquery for another aggregation:
SELECT v.date, v.NumVisits, COALESCE(fv.FirstVisit, 0) as NumFirstVisit
FROM (SELECT date, count(*) as NumVisits
FROM user_x_date
GROUP BY date
) v LEFT JOIN
(SELECT date, count(*) as FirstVisit
FROM (SELECT user_id, MIN(date) as date
FROM user_x_date
GROUP BY user_id
) x
GROUP BY date
) fv
ON v.date = fv.date;

How can i get count of customers per day by unique and repeat customer for specific date?

I am trying to get a result from my order table to get list of counts of customers who 1st time ordered and repeat orders. Something like below.
Date 1st time time repeat order
2014-09-01 43 90
2014-09-02 3 45
2014-09-03 12 30
2014-09-04 32 0
2014-09-05 1 98
I am beginner in sql and i ma using mysql.
My table structure is like.
OrderNumber int
OrderDate datetime
CustomerID int
I have tried this query in mysql but it only gives me first timed ordered count.
SELECT DATE(OrderDate), COUNT(*)
FROM orders T JOIN (
SELECT MIN(OrderDate) as minDate, CustomerID
FROM orders
GROUP BY CustomerID) T2 ON T.OrderDate = T2.minDate AnD T.CustomerID = T2.CustomerID
GROUP BY DATE(T.OrderDate)
You can get the total orders per day by grouping on OrderDate:
SELECT OrderDate, COUNT(OrderNumber) AS total FROM orders GROUP BY OrderDate
And you can get the no. of first orders per day from the following query :
SELECT OrderDate, COUNT(q1.CustomerID) AS first FROM (SELECT CustomerID, min(OrderDate) AS OrderDate FROM orders GROUP BY CustomerID)q1 GROUP BY q1.OrderDate
Now join these two on OrderDate to get the distribution of first and repeated orders :
SELECT a.OrderDate, a.first, (b.total - a.first) AS repeated FROM
(SELECT OrderDate, COUNT(q1.CustomerID) AS first FROM (SELECT CustomerID, min(OrderDate) AS OrderDate FROM orders GROUP BY CustomerID)q1 GROUP BY q1.OrderDate)a
JOIN
(SELECT OrderDate, COUNT(OrderNumber) AS total FROM orders GROUP BY OrderDate)b
on(a.OrderDate = b.OrderDate)
A slightly complicated query but this should do:
First Time Users: Just Group by customerID to get the min orderdate and then group by on that date to get the number of new users on a particular day. Query would look like this:
select date(mdate) as day, COUNT(*) from (select customerid, min(orderdate) as mDate from orders GROUP BY CustomerID)q1 GROUP BY day;
Repeat Users: First filter out all such orderno which were placed as first orders and then do a group by orderdate to get repeat. Query would be :
select date(orderdate) day, COUNT(*) from (select * from orders where orderno not in (select orders.orderno from orders JOIN (select customerid, min(orderdate) as mdate from orders GROUP BY CustomerID)as order2 ON (orders.customerid = order2.customerid) and (orders.orderdate = order2.mdate))) as q1 GROUP BY day;
You can do a join on day for both these queries to get combined results in a way you mentioned. Let me know if doesn't work
EDIT:
This would be the complete query: Here I am doing a UNION on both left and right outer joins since it might happen that you come across where there are no new requests or no repeated requests. This would take care of both the scenarios.
select q2.*, q3.repeated from (select date(mdate) as day, COUNT(*) as first from (select customerid, min(orderdate) as mDate from orders GROUP BY CustomerID)q1 GROUP BY day) as q2 LEFT OUTER JOIN (select date(orderdate) day, COUNT(*) as repeated from (select * from orders where orderno not in (select orders.orderno from orders JOIN (select customerid, min(orderdate) as mdate from orders GROUP BY CustomerID)as order2 ON (orders.customerid = order2.customerid) and (orders.orderdate = order2.mdate))) as q1 GROUP BY day) as q3 on q2.day = q3.day UNION select q2.*, q3.repeated from (select date(mdate) as day, COUNT(*) as first from (select customerid, min(orderdate) as mDate from orders GROUP BY CustomerID)q1 GROUP BY day) as q2 RIGHT OUTER JOIN (select date(orderdate) day, COUNT(*) as repeated from (select * from orders where orderno not in (select orders.orderno from orders JOIN (select customerid, min(orderdate) as mdate from orders GROUP BY CustomerID)as order2 ON (orders.customerid = order2.customerid) and (orders.orderdate = order2.mdate))) as q1 GROUP BY day) as q3 on q2.day = q3.day
this is my answer but not sure is still can improve.
SELECT userID, COUNT(*) AS repeat_order_cnt FROM
(SELECT DATE(OrderDate) AS order_DT, userID, COUNT(*) AS no_of_order FROM order
AND YEAR(orderDate) = '2015'
AND MONTH(orderDate) = '01'
GROUP BY order_DT,userID) AS order2
GROUP BY userID
HAVING COUNT(*) > 1

least value in count

i have a table employee(id,dept_id,salary,hire_date,job_id) . the following query i have to execute.
Show all the employee who were hired on the day of the week on which least no of employee were hired.
i have done the query, but am not able to get the least. please check if am correct.
select id, WEEKDAY(hire_date)+1 as days,count(WEEKDAY(hire_date)+1) as count
from test.employee group by days
This should get you the weekday on which the least number of employees were hired:
SELECT
count(id) as `Total`,
WEEKDAY(hire_date) as `DoW`
FROM
test.employee
GROUP BY `DoW`
ORDER BY `Total` DESC LIMIT 1;
select id from test.employee where hire_date in
( select count(id) count,hire_date
from test.employee
order by count desc
limit 1)
this should work
You may try this, as it will not limit to one record if you have multiple week days where the same least number of employees were hired. In reality it makes sense. The following is based on sample data.
Query:
-- find minimum id count
SELECT MIN(e.counts) INTO #min
FROM (SELECT COUNT(*) as counts,
WEEKDAY(hire_date+1) as day
FROM employee
GROUP BY WEEKDAY(hire_date+1)) e
;
-- show weekdays with minimum id counts
SELECT e2.counts as mincount,
WEEKDAY(e1.hire_date+1) as weekday
FROM employee e1
JOIN (SELECT COUNT(id) as counts,
WEEKDAY(hire_date+1) as day
FROM employee
GROUP BY day
HAVING COUNT(*) = #min) e2
ON WEEKDAY(e1.hire_date+1) = e2.day;
Results:
MINCOUNT WEEKDAY
1 6
1 3
1 4
1 2
SQLFIDDLE
select min(id), WEEKDAY(hire_date)+1 as days,count(WEEKDAY(hire_date)+1) as count
from test.employee group by days
SELECT
*
FROM
employee
WHERE
DAYOFWEEK(hire_date)
IN
(
SELECT
weekday
FROM
(
SELECT
count(*) as bcount,
DAYOFWEEK(hire_date) as weekday
FROM
employee as a
GROUP BY
weekday
HAVING
bcount = (
SELECT
MIN(tcount)
FROM
(
SELECT
count(*) as tcount,
DAYOFWEEK(hire_date) as weekday
FROM
employee
GROUP BY
weekday
) as t
)
) as q