Difference between last two entries in series

[IHam]

Hi, I have the following data, which is a long list of percentile values in a time series which are particular to an identifier. I wish to create a measure which will show if a percentile value has dropped dramatically since the last entry for that item. So i need to create a measure which calculates the difference between the last "number time" and the previous one (this varies depending on the identifier), to see if this difference is below a cut off value.

So for instance it would return a value of 0.9 for 54310 and -10.5 for 54319

Any help would be great.

Thanks

Ian

Identifier	Percentile	Number Time
54310	26	38
54310	42.4	44
54310	33.3	46
54310	44.2	48
54319	28.1	37
54319	12.2	44
54319	17.5	46
54319	7	50
54335	14.9	37
54335	34.3	44
54335	17.5	46
54335	36.8	50

[AlB]

Hi @IHam

Does the 'Number Time' column determine the order you want to follow? Is it unique for a given identifier?

It seems so but I'd like confirmation. Then we could use it as a sort of index 

[IHam]

Hi, yes the number time is the number order, so 50 is a later entry than 46. Each identifier will have a unique number time value as well. Many thanks for any help in advance

[AlB · Accepted solution]

Hi @IHam

I assume that in your initial post the 0.9 for 54310 is a typo and should be 10.9 

You could set [Identifier] in the rows of a matrix and the measure below in values.

Let me know if this helps.

DeltaLastEntries =
VAR Last_Time =
    MAX ( Table1[Number Time] )
VAR Previous2Last_Time =
    CALCULATE ( MAX ( Table1[Number Time] ); Table1[Number Time] < Last_Time )
VAR Last_Percentile =
    CALCULATE (
        SELECTEDVALUE ( Table1[Percentile] );
        Table1[Number Time] = Last_Time
    )
VAR Previous2Last_Percentile =
    CALCULATE (
        SELECTEDVALUE ( Table1[Percentile] );
        Table1[Number Time] = Previous2Last_Time
    )
RETURN
    Last_Percentile - Previous2Last_Percentile

[IHam]

That's perfect, many thanks for all your efforts and expertise - most appreciated!

[AlB]

@IHam

No worries. glad it helped.

Perhaps some kudos then?