I have a Table Users (it has millions of rows)
Id Name Country Product
+----+---------------+---------------+--------------+
1 John Canada
2 Kate Argentina
3 Mark China
4 Max Canada
5 Sam Argentina
6 Stacy China
...
1000 Ken Canada
I want to fill the Product column with A, B or C based on percentages.
I have another table called CountriesStats like the following
Id Country A B C
+-----+---------------+--------------+-------------+----------+
1 Canada 60 20 20
2 Argentina 35 45 20
3 China 40 10 50
This table holds the percentage of people with each product. For example in Canada 60% of people have product A, 20% have product B and 20% have product C.
I would like to fill the Users table with data based on the Percentages in the second data. So for example if there are 1 million user in canada, I would like to fill 600000 of the Product column in the Users table with A 200000 with B and 200000 with C
Thanks for any help on how to do that. I do not mind doing it in multiple steps I jsut need hints on how can I achieve that in SQL
The logic behind this is not too difficult. Assign a sequential counter to each person in each country. Then, using this value, assign the correct product based on this value. For instance, in your example, when the number is less than or equal to 600,000 then 'A' gets assigned. For 600,001 to 800,000 then 'B', and finally 'C' to the rest.
The following SQL accomplishes this:
with toupdate as (
select u.*,
row_number() over (partition by country order by newid()) as seqnum,
count(*) over (partition by country) as tot
from users u
)
update u
set product = (case when seqnum <= tot * A / 100 then 'A'
when seqnum <= tot * (A + B) / 100 then 'B'
else 'C'
end)
from toupdate u join
CountriesStats cs
on u.country = cs.country;
The with statement defines an updatable subquery with the sequence number and total for each each country, on each row. This is a nice feature of SQL Server, but is not supported in all databases.
The from statement is joining back to the CountriesStats table to get the needed values for each country. And the case statement does the necessary logic.
Note that the sequential number is assigned randomly, using newid(), so the products should be assigned randomly through the initial table.
Related
I have this table. The table just a small example and has more obs.
id
CATEGORY
AMOUNT
1
TECH
120
1
FUN
220
2
OTHER
340
2
PARENTS
220
made by id category amount spent in each category.I want to select ID and Category in which the ID spents the most but in case if category is OTHER I want to get 2nd most spending category.
I have a constraint. I CANNOT use the the subquery and select with filter WHERE CATEGORY <> 'OTHER'. It just makes my machine to go out of the memory (For reasons Idk)
This is what I have tried.
I have tried to create a row_number () over (partition by id order by amount desc) rn.
and then
select id, category from table where row num = 1 group by 1,2
**buttt. I don't know how to say to query. If CATEGORY is OTHER then take row num=2 . **
id
CATEGORY
AMOUNT
ROW NUM
1
TECH
120
2
1
FUN
220
1
2
OTHER
340
1
2
PARENTS
220
2
Another thing I was thinking to do is to write qualify function
QUALIFY ROW_NUMBER() OVER (PARTITION BY ID ORDER BY AMOUNT DESC) <1.
Also here I am getting only 1st records in which there is also OTHER. If I could filter it out within QUALIFY and say if CATEGORY is 'OTHER' don't consider it.
I am using Databricks.
Suppose I have a table which stores data for students and the respected grades per classes. In most cases, there exists multiple rows per student in the table.
student_id math chemistry science
100 A B C <--------
100 B A D
100 D F C
200 B A C <--------
300 C D F <--------
300 A A A
400 F C B <--------
400 B A C
500 A B A <--------
I want to retrieve the first row as per the student_id as explained below.
Requested in postgreSQL:
student_id math chemistry science
100 A B C
200 B A C
300 C D F
400 F C B
500 A B A
At the point you decide on an ordering strategy (such as adding a column that records the date they took the exams and defining "first" as "most recent") you can get the "first" row with something like
SELECT *
FROM(
SELECT *, ROW_NUMBER() OVER(PARTITION BY student_id ORDER BY date_taken DESC) rn
FROM yourtable
) x
WHERE rn = 1
The ROW_NUMBER creates an incrementing counter 1,2,3,4.. that restarts from 1 for every different student id, and the rows are numbered in order of descending date (so most recent gets 1). By then requiring rn to be 1 we get the most recent row
You might decide to give the student their best marks, and maybe we could use the ascii value of the score (A is 65, B is 66 etc). If we add the scores up then the lowest total (ie order by ascending) is the best set of marks (BBB is better than AAF)
OVER(PARTITION BY student_id ORDER BY ASCII(math)+ASCII(chemistry)+ASCII(science))
This can be achieved with the following query:
SELECT DISTINCT ON (student_id) student_id, math, chemistry, science
FROM
student
ORDER BY
student_id
This query will return just a single row per student_id. You should determine how you want to actually order it (such as an index or timestamp), otherwise you can't guarantee that you get the same output each time.
But as a basic solution, you can use this to just get a single row if you don't care about the actual order and only care about removing duplicates.
The basic solution is - using MIN/MAX function to aggregate values:
SELECT student_id, MAX(math) math, MAX(chemistry) chemistry, MAX(science) science
FROM
student
GROUP BY
student_id
How do we eliminate duplicates by only selecting those with values in a certain field using with clause statement?
Query is something like this:
with x as (--queries with multiple join tables, etc.)
select distinct * from x
Output below:
Com_no Company Loc Rewards
1 Mccin India 50
1 Mccin India
2 Rowle China 18
3 Draxel China 11
3 Draxel China
4 Robo UK
As you can see, I get duplicate records. I want to get rid of the null values that are NOT unique. That is to say, Robo is unique since it only has 1 record with a null value in Rewards, so I want to keep that.
I tried this:
with x as (--queries with multiple join tables, etc.)
select distinct * from x where Rewards is not null
And of course that wasn't right, since it also got rid of 4 Robo UK
Expected output should be:
1 Mccin India 50
2 Rowle China 18
3 Draxel China 11
4 Robo UK
The problem is you're calling those rows duplicates, but they're not duplicates. They're different. So what you want to do is exclude rows where Rewards is null UNLESS there aren't any rows with a not null value, and then select the distinct rows. So something like:
select distinct *
from x a
where Rewards is not null
or (Rewards is null and not exists (select 1 from x b where a.Com_no = b.Com_no
and b.Rewards is not null)
Now your Robo row will still be included as there isn't a row in x for Robo where Rewards is not null, but the rows for the other Companies with null Rewards will be excluded as there are not null rows for them.
This is a prioritization query. One method is to use row_number(). If you want only one value per Com_no/Company/Loc, then:
select x.*
from (select x.*,
row_number() over (partition by Com_no, Company, Loc order by Rewards nulls last) as seqnum
from x
) x
where seqnum = 1;
Or even:
select Com_no, Company, Loc, max(Rewards)
from x
group by Com_no, Company, Loc;
I would like to ignore some of the results of my query as for all intents and purposes, some of the results are a duplicate, but based on the way the request was made, we need to use this hierarchy and although we are seeing different 'Company_Name' 's, we need to ignore one of the results.
Query:
SELECT
COUNT(DISTINCT A12.Company_name) AS Customer_Name_Count,
Company_Name,
SUM(Total_Sales) AS Total_Sales
FROM
some_table AS A12
GROUP BY
2
ORDER BY
3 ASC, 2 ASC
This code omits half a doze joins and where statements that are not germane to this question.
Results:
Customer_Name_Count Company_Name Total_Sales
-------------------------------------------------------------
1 3 Blockbuster 1,000
2 6 Jimmy's Bar 1,500
3 6 Jimmy's Restaurant 1,500
4 9 Impala Hotel 2,000
5 12 Sports Drink 2,500
In the above set, we can see that numbers 2 & 3 have the same count and the same total_sales number and similar company names. Is there a way to create a case statement that takes these 3 factors into consideration and then drops one or the other for Jimmy's enterprises? The other issue is that this has to be variable as there are other instances where this happens. And I would only want this to happen if the count and sales number match each other with a similar name in the company name.
Desired result:
Customer_Name_Count Company_Name Total_Sales
--------------------------------------------------------------
1 3 Blockbuster 1,000
2 6 Jimmy's Bar 1,500
3 9 Impala Hotel 2,000
4 12 Sports Drink 2,500
Looks like other answers are accurate based on assumption that Company_IDs are the same for both.
If Company_IDs are different for both Jimmy's Bar and Jimmy's Restaurant then you can use something like this. I suggest you get functional users involved and do some data clean-up else you'll be maintaining this every time this issue arise:
SELECT
COUNT(DISTINCT CASE
WHEN A12.Company_Name = 'Name2' THEN 'Name1'
ELSE A12.Company_Name
END) AS Customer_Name_Count
,CASE
WHEN A12.Company_Name = 'Name2' THEN 'Name1'
ELSE A12.Company_Name
END AS Company_Name
,SUM(A12.Total_Sales) AS Total_Sales
FROM some_table er
GROUP BY CASE
WHEN A12.Company_Name = 'Name2' THEN 'Name1'
ELSE A12.Company_Name
END
Your problem is that the joins you are using are multiplying the number of rows. Somewhere along the way, multiple names are associated with exactly the same entity (which is why the numbers are the same). You can fix this by aggregating by the right id:
SELECT COUNT(DISTINCT A12.Company_name) AS Customer_Name_Count,
MAX(Company_Name) as Company_Name,
SUM(Total_Sales) AS Total_Sales
FROM some_table AS A12
GROUP BY Company_id -- I'm guessing the column is something like this
ORDER BY 3 ASC, 2 ASC;
This might actually overstate the sales (I don't know). Better would be fixing the join so it only returned one name. One possibility is that it is a type-2 dimension, meaning that there is a time component for values that change over time. You may need to restrict the join to a single time period.
You need to have function to return a common name for the companies and then use DISTINCT:
SELECT DISTINCT
Customer_Name_Count,
dbo.GetCommonName(Company_Name) as Company_Name,
Total_Sales
FROM dbo.theTable
You can try to use ROW_NUMBER with window function to make row number by Customer_Name_Count and Total_Sales then get rn = 1
SELECT * FROM (
SELECT *,ROW_NUMBER() OVER(PARTITION BY Customer_Name_Count,Total_Sales ORDER BY Company_Name) rn
FROM (
SELECT
COUNT(DISTINCT A12.Company_name) AS Customer_Name_Count,
Company_Name,
SUM(Total_Sales) AS Total_Sales
FROM
some_table AS A12
GROUP BY
Company_Name
)t1
)t1
WHERE rn = 1
I have three columns, all consisting of 1's and 0's. For each of these columns, how can I calculate the percentage of people (one person is one row/ id) who have a 1 in the first column and a 1 in the second or third column in oracle SQL?
For instance:
id marketing_campaign personal_campaign sales
1 1 0 0
2 1 1 0
1 0 1 1
4 0 0 1
So in this case, of all the people who were subjected to a marketing_campaign, 50 percent were subjected to a personal campaign as well, but zero percent is present in sales (no one bought anything).
Ultimately, I want to find out the order in which people get to the sales moment. Do they first go from marketing campaign to a personal campaign and then to sales, or do they buy anyway regardless of these channels.
This is a fictional example, so I realize that in this example there are many other ways to do this, but I hope anyone can help!
The outcome that I'm looking for is something like this:
percentage marketing_campaign/ personal campaign = 50 %
percentage marketing_campaign/sales = 0%
etc (for all the three column combinations)
Use count, sum and case expressions, together with basic arithmetic operators +,/,*
COUNT(*) gives a total count of people in the table
SUM(column) gives a sum of 1 in given column
case expressions make possible to implement more complex conditions
The common pattern is X / COUNT(*) * 100 which is used to calculate a percent of given value ( val / total * 100% )
An example:
SELECT
-- percentage of people that have 1 in marketing_campaign column
SUM( marketing_campaign ) / COUNT(*) * 100 As marketing_campaign_percent,
-- percentage of people that have 1 in sales column
SUM( sales ) / COUNT(*) * 100 As sales_percent,
-- complex condition:
-- percentage of people (one person is one row/ id) who have a 1
-- in the first column and a 1 in the second or third column
COUNT(
CASE WHEN marketing_campaign = 1
AND ( personal_campaign = 1 OR sales = 1 )
THEN 1 END
) / COUNT(*) * 100 As complex_condition_percent
FROM table;
You can get your percentages like this :
SELECT COUNT(*),
ROUND(100*(SUM(personal_campaign) / sum(count(*)) over ()),2) perc_personal_campaign,
ROUND(100*(SUM(sales) / sum(count(*)) over ()),2) perc_sales
FROM (
SELECT ID,
CASE
WHEN SUM(personal_campaign) > 0 THEN 1
ELSE 0
end AS personal_campaign,
CASE
WHEN SUM(sales) > 0 THEN 1
ELSE 0
end AS sales
FROM the_table
WHERE ID IN
(SELECT ID FROM the_table WHERE marketing_campaign = 1)
GROUP BY ID
)
I have a bit overcomplicated things because your data is still unclear to me. The subquery ensures that all duplicates are cleaned up and that you only have for each person a 1 or 0 in marketing_campaign and sales
About your second question :
Ultimately, I want to find out the order in which people get to the
sales moment. Do they first go from marketing campaign to a personal
campaign and then to sales, or do they buy anyway regardless of these
channels.
This is impossible to do in this state because you don't have in your table, either :
a unique row identifier that would keep the order in which the rows were inserted
a timestamp column that would tell when the rows were inserted.
Without this, the order of rows returned from your table will be unpredictable, or if you prefer, pure random.