Slicing account balance data in BigQuery to generate a debit report - google-bigquery

I have a collection of account balances over time:
+-----------------+------------+-------------+-----------------------+
| account_balance | department | customer_id | timestamp |
+-----------------+------------+-------------+-----------------------+
| 5 | A | 1 | 2019-02-12T00:00:00 |
| -10 | A | 1 | 2019-02-13T00:00:00 |
| -35 | A | 1 | 2019-02-14T00:00:00 |
| 20 | A | 1 | 2019-02-15T00:00:00 |
+-----------------+------------+-------------+-----------------------+
Each record shows the total account balance of a customer at a specified timestamp. The account balance increases e.g. to 20 from -35, when a customer tops-up his account with 55. As a customer uses a services, his account balances decreases e.g. from 5 to -10.
I want to aggregate this data in two ways:
1) Get the debit, credit and balance (credit-debit) of a department per month and year. The results from April should be a summary of all previous months:
+---------+--------+-------+------------+-------+--------+
| balance | credit | debit | department | month | year |
+---------+--------+-------+------------+-------+--------+
| 5 | 10 | -5 | A | 1 | 2019 |
| 20 | 32 | -12 | A | 2 | 2019 |
| 35 | 52 | -17 | A | 3 | 2019 |
| 51 | 70 | -19 | A | 4 | 2019 |
+---------+--------+-------+------------+-------+--------+
A customer's account balance might not change every month. There might be account balance records of customer 1 in February, but not March.
Notes towards the solution:
use EXTRACT(MONTH from timestamp) month
use EXTRACT(YEAR from timestamp) year
GROUP BY month, year, department
2) Get the change of debit, credit and balance of a department by date.
+---------+--------+-------+------------+-------------+
| balance | credit | debit | department | date |
+---------+--------+-------+------------+-------------+
| 5 | 10 | -5 | A | 2019-01-15 |
| 15 | 22 | -7 | A | 2019-02-15 |
| 15 | 20 | -5 | A | 2019-03-15 |
| 16 | 18 | -2 | A | 2019-04-15 |
+---------+--------+-------+------------+-------------+
51 70 -19
When I create a SUM of the deltas, I should get the same values as the last row from results in 1).
Notes towards the solution:
use account_balance - LAG(account_balance) OVER(PARTITION BY department ORDER BY timestamp ASC) delta to compute deltas

Your question is unclear, but it sounds like you want to get the outstanding balance at any given point in time.
The following query does this for 1 point in time.
with calendar as (
select cast('2019-06-01' as timestamp) as balance_calc_ts
),
most_recent_balance as (
select customer_id, balance_calc_ts,max(timestamp) as most_recent_balance_ts
from <table>
cross join calendar
where timestamp < balance_calc_ts -- or <=
group by 1,2
)
select t.customer_id, t.account_balance, mrb.balance_calc_ts
from <table> t
inner join most_recent_balance mrb on t.customer_id = mrb.customer_id and t.timestamp = mrb.balance_calc_ts
If you need to calculate it at a series of points in time, you will need to modify the calendar CTE to return more dates. This is the beauty of CROSS JOINS in BQ!

Related

SQLite: generating customer counts for a date range (months) using a normalized table

I have a sales funnel dataset in SQLite and each row represents a movement through the funnel. As there are quite a few ways a potential customer can move through the funnel (and possibly even go backwards), I wasn't planning on flattening/denormalizing the table. How could I calculate "the number of customers per month up to today"?
customer | opp_value | status_old | status_new | current_status | status_change_date | current_lead_status | lead_created_date
cust_8 | 22 | confirmed | paying | paying | 2020-01-01 | Customer | 2020-01-01
cust_9 | 23 | confirmed | paying | churned | 2020-01-03 | Customer | 2020-01-02
cust_9 | 23 | paying | churned | churned | 2020-03-24 | Customer | 2020-02-25
cust_13 | 30 | negotiation | lost | paying | 2020-04-03 | Lost | 2020-03-20
cust_14 | 45 | qualified | confirmed | paying | 2020-03-03 | Customer | 2020-02-28
cust_14 | 45 | confirmed | paying | paying | 2020-04-03 | Customer | 2020-02-28
... | ... | ... | ... | ... | ... | ... | ...
We're assuming we use end-of-month as definition for whether a customer is still with us.
The result, with the above data should be:
month | customers
Jan-2020 | 2 (cust_8, cust_9)
Feb-2020 | 1 (cust_8, cust_9)
Mar-2020 | 1 (cust_8) # cust_9 churned
Apr-2020 | 2 (cust_8, cust_14)
May-2020 | 2 (cust_8, cust_14)
The part I'd really like to understand is how to create the month column, as I can't rely on the dates of status_change_date as there might be missing records. Would one have to manually generate that column? I know I can generate dates manually using:
WITH RECURSIVE cnt (
x
) AS (
SELECT 0
UNION ALL
SELECT x + 1
FROM cnt
LIMIT (
SELECT
ROUND(((julianday ('2020-05-01') - julianday ('2020-01-01')) / 30) + 1))
)
SELECT
date(julianday ('2020-01-01'), '+' || x || ' month') AS month
FROM cnt
but wondering if there is a better way? Would it possibly be easier to create a snapshot table and generate the current state of each customer for each date?
If you have the dates, you can use a brute-force method. This determines the most recent status for each customer for each date:
select d.date,
sum(as_of_status = 'paying')
from (select distinct d.date, t.customer,
first_value(status_new) over (partition by d.date, t.customer order by t. status_change_date desc) as as_of_status
from dates d join
t
on t.status_change_date <= d.date
) dc
group by d.date
order by d.date;

How do I use a historic value as at a particular month when there are no values for the given month?

I have 2 SQL Server tables.
PurchaseOrderReceivingLine (PORL) is a table that contains every receipt from a purchase order. This has hundreds of entries per month.
PartyRelationshipScore (PRS) is a table with a party (supplier) reference number (that is used to join to the PORL table) and a score out of 10 for relationship and price. It also has a date field for when the score is updated so we have a history of the updates.
What I want to achieve is a supplier summary for each month. So I would have Supplier #, TotalValue, LateParts etc. I'm fine with creating the code for that. What I'm struggling with is getting the score for the given month if there are no values for that month.
So, for example I might have a value of 5 on the 1st August. Then it doesn't change until the 1st October when it is increased to 6.
On the grouping, September will have a TotalValue & a LateParts value but because there are no records in September in the PRS table, it will return a NULL value. I need it to get the last value recorded and return that (in this case August's 5). So it will return;
Aug 2019 - 5
Sep 2019 - 5
Oct 2019 - 6
Thanks in advance.
PORL Table
+-------+----------------+-------+-------+
| PORL# | Date (UK) | Value | Party |
+-------+----------------+-------+-------+
| 1 | 1/8/2019 | 100 | 6 |
| 2 | 1/8/2019 | 250 | 6 |
| 3 | 1/9/2019 | 1000 | 6 |
| 4 | 1/10/2019 | 2000 | 6 |
+-------+----------------+-------+-------+
PRS Table
+-------------+------------+-------------------+------------+
| DateChanged (UK) | Party | RelationShipScore | PriceScore |
+-------------+------------+-------------------+------------+
| 1/8/2019 | 6 | 5 | 5 |
| 1/10/2019 | 6 | 6 | 7 |
+------------------+-------+-------------------+------------+
Preferred outcome
+----------+-------+------+------------+-------------------+------------+
| Supplier | Month | Year | TotalValue | RelationshipScore | PriceScore |
+----------+-------+------+------------+-------------------+------------+
| 6 | 8 | 2019 | 350 | 5 | 5 |
| 6 | 9 | 2019 | 1000 | 5 | 5 |
| 6 | 10 | 2019 | 2000 | 6 | 7 |
+----------+-------+------+------------+-------------------+------------+
The relationshipscore & pricescore for month 9 are based on it not changing from month 8.
I think this helps
select Supplier = T.Party
, Month = DATEPART(MONTH,T.[Date])
, Year = DATEPART(YEAR,T.[Date])
, T.TotalValue
, R.RelationShipScore
, R.PriceScore
from ( Select P.[Party],P.[Date],[TotalValue] = sum(P.[Value])
from PurchaseOrderReceivingLine P
group by P.[Party],P.[Date] ) T
outer apply ( select top 1 RelationShipScore , PriceScore
from PartyRelationshipScore
where Party = T.Party
and DateChanged <= T.[Date]
Order by DateChanged desc ) R

SQL: Get an aggregate (SUM) of a calculation of two fields (DATEDIFF) that has conditional logic (CASE WHEN)

I have a dataset that includes a bunch of stay data (at a hotel). Each row contains a start date and an end date, but no duration field. I need to get a sum of the durations.
Sample Data:
| Stay ID | Client ID | Start Date | End Date |
| 1 | 38 | 01/01/2018 | 01/31/2019 |
| 2 | 16 | 01/03/2019 | 01/07/2019 |
| 3 | 27 | 01/10/2019 | 01/12/2019 |
| 4 | 27 | 05/15/2019 | NULL |
| 5 | 38 | 05/17/2019 | NULL |
There are some added complications:
I am using Crystal Reports and this is a SQL Expression, which obeys slightly different rules. Basically, it returns a single scalar value. Here is some more info: http://www.cogniza.com/wordpress/2005/11/07/crystal-reports-using-sql-expression-fields/
Sometimes, the end date field is blank (they haven't booked out yet). If blank, I would like to replace it with the current timestamp.
I only want to count nights that have occurred in the past year. If the start date of a given stay is more than a year ago, I need to adjust it.
I need to get a sum by Client ID
I'm not actually any good at SQL so all I have is guesswork.
The proper syntax for a Crystal Reports SQL Expression is something like this:
(
SELECT (CASE
WHEN StayDateStart < DATEADD(year,-1,CURRENT_TIMESTAMP) THEN DATEDIFF(day,DATEADD(year,-1,CURRENT_TIMESTAMP),ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
ELSE DATEDIFF(day,StayDateStart,ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
END)
)
And that's giving me the correct value for a single row, if I wanted to do this:
| Stay ID | Client ID | Start Date | End Date | Duration |
| 1 | 38 | 01/01/2018 | 01/31/2019 | 210 | // only days since June 4 2018 are counted
| 2 | 16 | 01/03/2019 | 01/07/2019 | 4 |
| 3 | 27 | 01/10/2019 | 01/12/2019 | 2 |
| 4 | 27 | 05/15/2019 | NULL | 21 |
| 5 | 38 | 05/17/2019 | NULL | 19 |
But I want to get the SUM of Duration per client, so I want this:
| Stay ID | Client ID | Start Date | End Date | Duration |
| 1 | 38 | 01/01/2018 | 01/31/2019 | 229 | // 210+19
| 2 | 16 | 01/03/2019 | 01/07/2019 | 4 |
| 3 | 27 | 01/10/2019 | 01/12/2019 | 23 | // 2+21
| 4 | 27 | 05/15/2019 | NULL | 23 |
| 5 | 38 | 05/17/2019 | NULL | 229 |
I've tried to just wrap a SUM() around my CASE but that doesn't work:
(
SELECT SUM(CASE
WHEN StayDateStart < DATEADD(year,-1,CURRENT_TIMESTAMP) THEN DATEDIFF(day,DATEADD(year,-1,CURRENT_TIMESTAMP),ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
ELSE DATEDIFF(day,StayDateStart,ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
END)
)
It gives me an error that the StayDateEnd is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause. But I don't even know what that means, so I'm not sure how to troubleshoot, or where to go from here. And then the next step is to get the SUM by Client ID.
Any help would be greatly appreciated!
Although the explanation and data set are almost impossible to match, I think this is an approximation to what you want.
declare #your_data table (StayId int, ClientId int, StartDate date, EndDate date)
insert into #your_data values
(1,38,'2018-01-01','2019-01-31'),
(2,16,'2019-01-03','2019-01-07'),
(3,27,'2019-01-10','2019-01-12'),
(4,27,'2019-05-15',NULL),
(5,38,'2019-05-17',NULL)
;with data as (
select *,
datediff(day,
case
when datediff(day,StartDate,getdate())>365 then dateadd(year,-1,getdate())
else StartDate
end,
isnull(EndDate,getdate())
) days
from #your_data
)
select *,
sum(days) over (partition by ClientId)
from data
https://rextester.com/HCKOR53440
You need a subquery for sum based on group by client_id and a join between you table the subquery eg:
select Stay_id, client_id, Start_date, End_date, t.sum_duration
from your_table
inner join (
select Client_id,
SUM(CASE
WHEN StayDateStart < DATEADD(year,-1,CURRENT_TIMESTAMP) THEN DATEDIFF(day,DATEADD(year,-1,CURRENT_TIMESTAMP),ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
ELSE DATEDIFF(day,StayDateStart,ISNULL(StayDateEnd,CURRENT_TIMESTAMP))
END) sum_duration
from your_table
group by Client_id
) t on t.Client_id = your_table.client_id

How does DATEADD work when joining the same table with itself?

I have a table with monthly production values.
Example:
Outdate | Prod Value | ID
2/28/19 | 110 | 4180
3/31/19 | 100 | 4180
4/30/19 | 90 | 4180
I also have a table that has monthly forecast values.
Example:
Forecast End Date | Forecast Value | ID
2/28/19 | 120 | 4180
3/31/19 | 105 | 4180
4/30/19 | 80 | 4180
I want to create a table that has a row that contains the ID, the Prod Value, the current month (example: March) forecast, the previous month forecast, the next month forecast.
What I want:
ID | Prod Value | Outdate | Current Forecast | Previous Forecast | Next Forecast
4180 | 100 | 3/31/19 | 105 | 120 | 80
The problem is that when I used DATEADD to bring in the specific value from the Forecast table for the previous month, random months are missing from my final values.
I've tried adding in another LEFT JOIN / INNER JOIN with the DateDimension table when adding in the Next Month and Previous Month forecast, but that either does not solve the problem or adds in too many rows.
My DateDimension table that has these columns: DateKey
Date, Day, DaySuffix, Weekday, WeekDayName, IsWeekend, IsHoliday, DOWInMonth, DayOfYear, WeekOfMonth, WeekOfYear, ISOWeekOfYear, Month, MonthName, Quarter, QuarterName, Year, MMYYYY, MonthYear, FirstDayOfMonth, LastDayOfMonth, FirstDayOfQuarter, LastDayOfQuarter, FirstDayOfYear, LastDayOfYear, FirstDayOfNextMonth, FirstDayOfNextYear
My query is along these lines (abbreviated for simplicity)
SELECT A.ArchiveKey, BH.ID, d.[Date], BH.Outdate, BH.ProdValue, BH.Forecast, BHP.Forecast, BHN.Foreceast
FROM dbo.BudgetHistory bh
INNER JOIN dbo.DateDimension d ON bh.outdate = d.lastdayofmonth
INNER JOIN dbo.Archive a ON bh.ArchiveKey = a.ArchiveKey
LEFT JOIN dbo.BudgetHistory bhp ON bh.ID = bhp.ID AND bhp.outdate = DATEADD(m, - 1, bh.Outdate)
LEFT JOIN dbo.BudgetHistory bhn ON bh.ID = bhn.ID AND bhn.outdate = DATEADD(m, 1, bh.Outdate)
WHERE bh.ID IS NOT NULL
I get something like this:
+------+------------+---------+------------------+-------------------+---------------+
| ID | Prod Value | Outdate | Current Forecast | Previous Forecast | Next Forecast |
+------+------------+---------+------------------+-------------------+---------------+
| 4180 | 110 | 2/28/19 | 120 | NULL | NULL |
| 4180 | 100 | 3/31/19 | 105 | 120 | 80 |
| 4180 | 90 | 4/30/19 | 80 | NULL | NULL |
+------+------------+---------+------------------+-------------------+---------------+
And the pattern doesn't seem to follow anything reasonable.
I want the values to be filled in for each row.
You could join the tables, then use window functions LEAD() and LAG() to recover the next and previous forecast values:
SELECT
p.ID,
p.ProdValue,
p.Outdate,
f.ForecastValue,
LAG(f.ForecastValue) OVER(PARTITION BY f.ID ORDER BY f.ForecastEndDate) PreviousForecast,
LEAD(f.ForecastValue) OVER(PARTITION BY f.ID ORDER BY f.ForecastEndDate) NextForecast
FROM prod p
INNER JOIN forecast f ON p.ID = f.ID AND p.Outdate = f.ForecastEndDate
This demo on DB Fiddle with your sample data returns:
ID | ProdValue | Outdate | ForecastValue | PreviousForecast | NextForecast
---: | --------: | :------------------ | ------------: | ---------------: | -----------:
4180 | 110 | 28/02/2019 00:00:00 | 120 | null | 105
4180 | 100 | 31/03/2019 00:00:00 | 105 | 120 | 80
4180 | 90 | 30/04/2019 00:00:00 | 80 | 105 | null
DATEADD only does end of month adjustments if the newly calculated value isn't a valid date. So DATEADD(month,-1,'20190331') produces 28th February. But DATEADD(month,-1,'20190228') produces 28th January, not the 31st.
I would probably go with GMB's answer. If you want to do something DATEADD based though, you can use:
bhp.outdate = DATEADD(month, DATEDIFF(month,'20010131', bh.Outdate) ,'20001231')
This always works out the last day of the previous month from bh.OutDate, but it does it by computing it as an offset from a fixed date, and then applying that offset to a different fixed date.
You can just reverse the places of 20010131 and 20001231 to compute the month after rather than the month before. There's no significance about them other than them both having 31 days and having the "one month apart" relationship we're wishing to apply.

SQL query to select today and previous day's price

I have historic stock price data that looks like the below. I want to generate a new table that has one row for each ticker with the most recent day's price and its previous day's price. What would be the best way to do this? My database is Postgres.
+---------+------------+------------+
| ticker | price | date |
+---------+------------+------------|
| AAPL | 6 | 10-23-2015 |
| AAPL | 5 | 10-22-2015 |
| AAPL | 4 | 10-21-2015 |
| AXP | 5 | 10-23-2015 |
| AXP | 3 | 10-22-2015 |
| AXP | 5 | 10-21-2015 |
+------- +-------------+------------+
You can do something like this:
with ranking as (
select ticker, price, dt,
rank() over (partition by ticker order by dt desc) as rank
from stocks
)
select * from ranking where rank in (1,2);
Example: http://sqlfiddle.com/#!15/e45ea/3
Results for your example will look like this:
| ticker | price | dt | rank |
|--------|-------|---------------------------|------|
| AAPL | 6 | October, 23 2015 00:00:00 | 1 |
| AAPL | 5 | October, 22 2015 00:00:00 | 2 |
| AXP | 5 | October, 23 2015 00:00:00 | 1 |
| AXP | 3 | October, 22 2015 00:00:00 | 2 |
If your table is large and have performance issues, use a where to restrict the data to last 30 days or so.
Best bet is to use a window function with an aggregated case statement which is used to create a pivot on the data.
You can see more on window functions here: http://www.postgresql.org/docs/current/static/tutorial-window.html
Below is a pseudo code version of where you may need to head to answer your question (sorry I couldn't validate it due to not have a postgres database setup).
Select
ticker,
SUM(CASE WHEN rank = 1 THEN price ELSE 0 END) today,
SUM(CASE WHEN rank = 2 THEN price ELSE 0 END) yesterday
FROM (
SELECT
ticker,
price,
date,
rank() OVER (PARTITION BY ticker ORDER BY date DESC) as rank
FROM your_table) p
WHERE rank in (1,2)
GROUP BY ticker.
Edit - Updated the case statement with an 'else'