Average date difference between two dates in same column

[Anonymous]

Hi everyone,

I want to calculate the average waiting time for our customers. The table shows a small sample set (see below). 

We only have data once a month. To be precise as possible, I want to return the number of days for 0,5 month, because we don't know what the exact waiting time is, it could be 1 day or 31 days. 

Notice that customer 4 has two periods of waiting time. How can we prevent that Power BI doesn't see this as one period, from 1-1-2018 to 1-7-2019, but sees it as two seperate periods?

The average waiting time is as follows:

Customer_ID       Waiting time in days

1                                 107,5

2                                 14,5

3                                 46

4                                 362,5

Average waiting time: 133

The order of date is European, so 1-10-2018 is October, 1st, 2018. For customer 1, the waiting period starts then. It ends at 1-2-2019, so 31 + 30 + 31 + 15,5 = 107,5 days.

Customer_ID	Date
1	1-10-2018
1	1-11-2018
1	1-12-2018
1	1-1-2019
1	1-2-2019
2	1-2-2020
3	1-5-2020
3	1-6-2020
4	1-1-2018
4	1-2-2018
4	1-3-2018
4	1-4-2018
4	1-5-2018
4	1-6-2018
4	1-1-2019
4	1-2-2019
4	1-3-2019
4	1-4-2019
4	1-5-2019
4	1-6-2019
4	1-7-2019

[v-juanli-msft · Accepted solution]

Hi @Anonymous 

Sorry for replying late.

Please create measures

lastmonth = CALCULATE(MAX('date'[year-month]),FILTER(ALLSELECTED('Table'),'Table'[CustomerID]=MAX('Table'[CustomerID])))

measure in days =
VAR days1 =
    CALCULATE (
        SUM ( 'date'[days_m] ),
        VALUES ( 'date'[year-month] ),
        FILTER (
            ALLSELECTED ( 'Table' ),
            'Table'[CustomerID]
                = MAX ( 'Table'[CustomerID] )
        )
    )
RETURN
    IF (
        [lastmonth]
            = MAX ( 'date'[year-month] )
            && [lastmonth]
                <> BLANK (),
        MAX ( 'date'[days_m] ) / 2,
        days1
    )


each days = SUMX(FILTER(ALLSELECTED('Table'),'Table'[CustomerID]=MAX('Table'[CustomerID])),[measure in days])

average dyas = SUMX(ALLSELECTED('Table'),[measure in days])/2

The date table used above

date =
ADDCOLUMNS (
    CALENDARAUTO (),
    "year", YEAR ( [Date] ),
    "month", MONTH ( [Date] ),
    "year-month", FORMAT (
        [Date],
        "yyyy-mm"
    )
)

add a column in date table
days_m = CALCULATE(COUNT('date'[Date]),ALLEXCEPT('date','date'[year-month]))

Best Regards
Maggie
Community Support Team _ Maggie Li
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[v-juanli-msft]

Hi @Anonymous 

I get results as below:

Will work further to get the final result as you expected.

Before this, please check the results of your customer 1, if i undertsand your rule correctly, the result of your example is incorrect.

Please let me know if i undertsand correctly.

Best Regards

Maggie

[Anonymous]

Hi @v-juanli-msft 

Thanks for your help, I really appreciate it. 

Unfortunately, that isn't the result I'm looking for. I created a sample set of the real data and with the outcomes the measure should give. I hope this helps. 

https://drive.google.com/file/d/1j0HxxkZddVkAIpJBjwuusiOVzVaeU7cF/view?usp=sharing

[v-juanli-msft · Accepted solution]

Hi @Anonymous 

Sorry for replying late.

Please create measures

lastmonth = CALCULATE(MAX('date'[year-month]),FILTER(ALLSELECTED('Table'),'Table'[CustomerID]=MAX('Table'[CustomerID])))

measure in days =
VAR days1 =
    CALCULATE (
        SUM ( 'date'[days_m] ),
        VALUES ( 'date'[year-month] ),
        FILTER (
            ALLSELECTED ( 'Table' ),
            'Table'[CustomerID]
                = MAX ( 'Table'[CustomerID] )
        )
    )
RETURN
    IF (
        [lastmonth]
            = MAX ( 'date'[year-month] )
            && [lastmonth]
                <> BLANK (),
        MAX ( 'date'[days_m] ) / 2,
        days1
    )


each days = SUMX(FILTER(ALLSELECTED('Table'),'Table'[CustomerID]=MAX('Table'[CustomerID])),[measure in days])

average dyas = SUMX(ALLSELECTED('Table'),[measure in days])/2

The date table used above

date =
ADDCOLUMNS (
    CALENDARAUTO (),
    "year", YEAR ( [Date] ),
    "month", MONTH ( [Date] ),
    "year-month", FORMAT (
        [Date],
        "yyyy-mm"
    )
)

add a column in date table
days_m = CALCULATE(COUNT('date'[Date]),ALLEXCEPT('date','date'[year-month]))

Best Regards
Maggie
Community Support Team _ Maggie Li
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[Anonymous]

Hi @v-juanli-msft ,

I was wondering if it is possible to add another specification in the measure. For example, if you add a column to specifiy the country for each customer, let's say Spain and France (see image). I tried to change the first measure:

Spain measure in days = var days1 = CALCULATE(SUM('date'[days_m]);'Table'[Country] = "Spain";

VALUES('date'[year-month]);FILTER(ALLSELECTED('Table');'Table'[CustomerID]=MAX('Table'[CustomerID]))) return IF([lastmonth]=MAX('date'[year-month])&&[lastmonth]<>BLANK();MAX('date'[days_m])/2;days1)

And secondly, altering the next measure:

Spain each day = SUMX(FILTER(ALLSELECTED('Table');'Table'[CustomerID]=MAX('Table'[CustomerID]));[Spain measure in days])

But that makes no difference compared to the measure 'each days'... Is there a way to create this measure?

[v-juanli-msft · Accepted solution]

Hi @Anonymous 

Is this ok?

lastmonth =
CALCULATE (
    MAX ( 'date'[year-month] ),
    FILTER (
        ALLSELECTED ( 'Table' ),
        'Table'[CustomerID]
            = MAX ( 'Table'[CustomerID] )
            && 'Table'[country]
                = MAX ( 'Table'[country] )
    )
)

measure in days =
VAR days1 =
    CALCULATE (
        SUM ( 'date'[days_m] ),
        VALUES ( 'date'[year-month] ),
        FILTER (
            ALLSELECTED ( 'Table' ),
            'Table'[CustomerID]
                = MAX ( 'Table'[CustomerID] )
                && 'Table'[country]
                    = MAX ( 'Table'[country] )
        )
    )
RETURN
    IF (
        [lastmonth]
            = MAX ( 'date'[year-month] )
            && [lastmonth]
                <> BLANK (),
        MAX ( 'date'[days_m] ) / 2,
        days1
    )


each days =
SUMX (
    FILTER (
        ALLSELECTED ( 'Table' ),
        'Table'[CustomerID]
            = MAX ( 'Table'[CustomerID] )
            && 'Table'[country]
                = MAX ( 'Table'[country] )
    ),
    [measure in days]
)


total eachdays =
IF (
    ISINSCOPE ( 'Table'[CustomerID] ),
    [each days],
    AVERAGEX (
        SUMMARIZE (
            'Table',
            'Table'[CustomerID],
            "m", [each days]
        ),
        [m]
    )
)

Best Regards

Maggie

[Anonymous]

@v-juanli-msft 

Thank you so much!

[Anonymous]

Hi @v-juanli-msft ,

That's no problem! I am responding late myself. 

I wanted to let you know that the measures worked. So, thank you very much for that. I did discover that the measure 'each days' didn't show the correct Total (it showed the number of the last row). This measures solves that:

each days_total =

IF(HASONEVALUE('Table'[CustomerID]);[each days];

AVERAGEX(

SUMMARIZE(

'Table';

[CustomerID];

"Measure";[each days]);

[Measure] ) )

[v-juanli-msft]

Hi @Anonymous 

For customer 2, he has only one date 2020/2/1,

for customer 3, he has a period from 2020/5/1-2020/6/1,

why the "waiting time in days" for them is 14.5, 46,

 For customer 4, if we say he has a period from 2018/1/1-2019/7/1, then his "waiting time in days" should be 500+days.

Please explain the calculation rule.

Best Regards

Maggie

[Anonymous]

Hi @v-juanli-msft 

Thanks for your reply. 

Customer 2 has only one date, because that customer wasn't anymore on the waiting list for 2020/3/1. But since we don't know when that customer wasn't anymore on the list on February, the number of waiting days is 14.5. This is our best approach/estimation we can make (29 days this year in Febrary, so 0.5 = 14.5 days). 

So for customer 3 is it as follows: 31 days in May, 15 days (0.5*30) in June = 46. 

Customer 4 appeared to be two times on the waiting list. 5.5 month in 2018 and 6.5 month in 2019. Between those two periods, the customer wasn't on the waiting list, hence those months should not be included. Therefore, the number of waiting days is 362.5. 

I hope this helps. 

[Greg_Deckler]

You might be able to use a variation of Cthulhu to get your group id column. https://community.powerbi.com/t5/Quick-Measures-Gallery/Cthulhu/m-p/509739#M211

[Greg_Deckler]

You will need some kind of group identifier column and then you could do something like MTBF. See my article on Mean Time Between Failure (MTBF) which uses EARLIER: http://community.powerbi.com/t5/Community-Blog/Mean-Time-Between-Failure-MTBF-and-Power-BI/ba-p/3395...