DATEDIFF between duplicate IDs

[Matt_Alexander]

Hi all,

Wondering if anyone could help me work out how to approach the following challenge:

I have a table with millions of IDs in one columns, and dates of when they were sold in another.

In the event that an ID is repeated, it means the product has been resold and I need to calculate the average datediff between those two dates for the repeated IDs.

Would the best approach be to create another table by performing a Group By on the IDs? If so how would I then use the datediff to only calculate between the two dates relating to the repeated IDs?

Many thanks for any direction on this one!

Matt

[Anonymous · Accepted solution]

So... we take the max(date sold) minus min(date sold) giving us the total # of days between sales, then divide by the number of sales (minus 1)... and that gives the average number of days between sales?   I think?

Jan 1, 2017

Jan 9, 2017

  ... 8 days between sales  -- Jan 9 - Jan 8 / (2 sales - 1) = 8

Jan 1, 2017

 Jan 2, 2017

Jan 9, 2017

  ... still 8 days, but now with another sale, we get average of 4 days.  The exact date off the "middle" sale doesn't matter, at least in my head!?    Currently Jan2-Jan1=1, Jan9-Jan2=7,  (7+1)/2=4 ?

Avg Days Between Sale := DIVIDE(MAX(MyTable[SaleDate]) - MIN(MyTable[SaleDate]), COUNTROWS(MyTable) - 1)

Maybe?! 🙂

[Anonymous]

What if the same ID comes up ... 3 times?  47 times?

[Matt_Alexander]

Hi @Anonymous

Good question! The time taken between each date (sale) should be included in the average calculation.

A product (the same ID) can be sold numerous times.

Let me know if you need any further information!

Thanks,

Matt

[Anonymous · Accepted solution]

So... we take the max(date sold) minus min(date sold) giving us the total # of days between sales, then divide by the number of sales (minus 1)... and that gives the average number of days between sales?   I think?

Jan 1, 2017

Jan 9, 2017

  ... 8 days between sales  -- Jan 9 - Jan 8 / (2 sales - 1) = 8

Jan 1, 2017

 Jan 2, 2017

Jan 9, 2017

  ... still 8 days, but now with another sale, we get average of 4 days.  The exact date off the "middle" sale doesn't matter, at least in my head!?    Currently Jan2-Jan1=1, Jan9-Jan2=7,  (7+1)/2=4 ?

Avg Days Between Sale := DIVIDE(MAX(MyTable[SaleDate]) - MIN(MyTable[SaleDate]), COUNTROWS(MyTable) - 1)

Maybe?! 🙂

[Matt_Alexander · Accepted solution]

Hi again,

Qucik update, it turns out that my technical indications were incorrect and the sales are not going to coming through this way.

However, intrigued by the suggestion I've had a quick look through but can't seem to get it to work:

ID	Date Sold	Time taken to sell
1	01/01/2017	0
1	03/01/2017	2
3	05/01/2017	0		Average Days to Sell
3	10/01/2017	5		3,5
	Jan10-Jan1 / 2 sales = 5?
	Jan10-Jan1 / (2-1) sales = 10?

Many thanks for your help anyway.

Matt

[Matt_Alexander]

Interesting approach, let me have a play around and see what the results are!

Thank you for the suggestion @Anonymous

[v-huizhn-msft]

Hi @Matt_Alexander,

If you still do not resolved your problem, you'd better post some sample table for further analysis. And welcome to share your workaround and mark the helpful reply as answer if you have resolved it.

Best Regards,
Angelia