De-cumulate values with dates and dimension

[Yggdrasill]

I have a table like so:

ID| Date| Value

15123	21.7.2010	121
15123	28.7.2010	211
15123	4.8.2010	299
15123	11.8.2010	386
15123	18.8.2010	463
15123	25.8.2010	588
15123	1.9.2010	645
15123	8.9.2010	704
15123	15.9.2010	766
15999	22.9.2010	802
15999	29.9.2010	946
15999	6.10.2010	991
15999	13.10.2010	1006
15999	20.10.2010	1021
15123	13.7.2011	2
15123	20.7.2011	8
15123	27.7.2011	27
15123	3.8.2011	41
15123	10.8.2011	110
15123	17.8.2011	139
15123	24.8.2011	219
15123	31.8.2011	275
15123	7.9.2011	312
15123	14.9.2011	373
15123	21.9.2011	404
15123	28.9.2011	428
15999	5.10.2011	451
15999	12.10.2011	473
15999	19.10.2011	476

The values from the source are already cumulative. 

My desired output is this:
ID | Date| Value| Decumulative values

15123	21.7.2010	121	121
15123	28.7.2010	211	90
15123	4.8.2010	299	88
15123	11.8.2010	386	87
15123	18.8.2010	463	77
15123	25.8.2010	588	125
15123	1.9.2010	645	57
15123	8.9.2010	704	59
15123	15.9.2010	766	62
15123	22.9.2010	802	36
15999	29.9.2010	946	946
15999	6.10.2010	991	45
15999	13.10.2010	1006	15
15999	20.10.2010	1021	15
15123	13.7.2011	2	2
15123	20.7.2011	8	6
15123	27.7.2011	27	19
15123	3.8.2011	41	14
15123	10.8.2011	110	69
15123	17.8.2011	139	29
15123	24.8.2011	219	80
15123	31.8.2011	275	56
15123	7.9.2011	312	37
15123	14.9.2011	373	61
15123	21.9.2011	404	31
15123	28.9.2011	428	24
15999	5.10.2011	451	451
15999	12.10.2011	473	22
15999	19.10.2011	476	3

So basically - De-cumulate the Cumulative values within each ID and YEAR

I'm trying to use EARLIER() but I'm kinda lost here

Appreciate your help 

[Yggdrasill · Accepted solution]

Ok I just solved this one. Maybe there's a better way

De-cumulative value =
VAR _predate =
    CALCULATE (
        MAX ( 'Table'[Date] );
        FILTER ( 'Table'; 'Table'[Date] < EARLIER ( 'Table'[Date] ) );
        'Table'[ID] = EARLIER ( 'Table'[ID] );
        YEAR ( 'Table'[Date] ) = YEAR ( EARLIER ( 'Table'[Date] ) )
    )
VAR _prevalue =
    CALCULATE (
        SUM ( 'Table'[CumulativeValue] );
        FILTER ( 'Table'; 'Table'[Date] = _predate );
        'Table'[ID] = EARLIER ( 'Table'[ID] )
    )
RETURN
    'Table'[CumulativeValue] - _prevalue

[Yggdrasill · Accepted solution]

Ok I just solved this one. Maybe there's a better way

De-cumulative value =
VAR _predate =
    CALCULATE (
        MAX ( 'Table'[Date] );
        FILTER ( 'Table'; 'Table'[Date] < EARLIER ( 'Table'[Date] ) );
        'Table'[ID] = EARLIER ( 'Table'[ID] );
        YEAR ( 'Table'[Date] ) = YEAR ( EARLIER ( 'Table'[Date] ) )
    )
VAR _prevalue =
    CALCULATE (
        SUM ( 'Table'[CumulativeValue] );
        FILTER ( 'Table'; 'Table'[Date] = _predate );
        'Table'[ID] = EARLIER ( 'Table'[ID] )
    )
RETURN
    'Table'[CumulativeValue] - _prevalue

[v-eachen-msft · Accepted solution]

Hi @Yggdrasill ,

I have another solution that needs an index column. Then use EARLIER() function to get previous row value. 

ValueDiff =
Table1[Value]
    - CALCULATE (
        FIRSTNONBLANK ( Table1[Value], 1 ),
        FILTER (
            Table1,
            Table1[Index]
                = EARLIER ( Table1[Index] ) - 1
                && Table1[ID] = EARLIER ( Table1[ID] )
        )
    )

You can try my method and consider which one is more acceptable.