Assume there's a table T1 with columns [state],[product],[price],[store],[date], and a time table T2. There's a relationship between T2's [pk_date] column and T1's [date] columnit.
and I create a measure [5days moving average price]:
calculate
(
average('T1'[price])
,
DATESINPERIOD
(
'T2'[pk_Date],
LASTDATE ( 'T2'[pk_Date] )-1,
-5,
DAY
)
)
but for certain state and product, is there anyway I can get a rank of stores based on the [5days moving average price]? I tried something like
:=RANKX(ALLSELECTED('T1'[store]),[5Days Moving AVERAGE PRICE],,1)
but it didn't work.
Related
I have a DAX formula that is performing really bad and hopefully someone here can suggest a solution.
I have a table that contains about 400000 rows of data. ProductID's (example field), startdate, enddate and an IsActive flag field. The data out of this table should be reported in several ways. In some reports I want to see all of the active products within a selected period of time and in other reports, I only want to see the number of products that were active on the last day of the month.
So, I have created two DAX queries to calculate this.
First I calculate the active products:
_Calc_Count Fields :=
CALCULATE (
DISTINCTCOUNT ( MyFactTable[ProductID] ),
FILTER (
MyFactTable,
MyFactTable[StartDate] <= CALCULATE ( MAX ( 'Date'[Date] ) )
&& MyFactTable[EndDate] >= CALCULATE ( MIN ( 'Date'[Date] ) )
),
MyFactTable[IsActive] = 1
)
Please be aware of the fact that the report this calculation is used in can also contain a date range (even a whole year (or multiple years) can be selected with a startdate and enddate selected in the filter). The report also slices on other filters like Client Group.
Then I have a second calculation that uses the first one and applies the LASTNONBLANK function:
Last Non Blank Value :=
CALCULATE (
[_Calc_Count Fields],
LASTNONBLANK ( 'Date'[Date], [_Calc_Count Fields] )
)
Both calculations are very, very slow.
Can anyone suggest a better approach? Can the DAX formula be optimized or should it completely be rewritten?
ps. I am using Analysis Services Tabular Model.
Thank you all in advance for your responses!
there are many points to consider for optimizing.
First of all, you need to understand where is the bottleneck.
I would do three separate preliminary tests:
A) change the DISTINCTCOUNT with a simple COUNT
B) Remove the FILTER
C) Remove the IsActive
Then you can understand where to prioritize your effort, however there are some very simple general optimization you can do anyway:
1.Make use of variables, therefore the formula becomes:
_Calc_Count Fields 3:=
VAR _startdate = CALCULATE ( MAX ( 'Date'[Date] ) )
VAR _enddate = CALCULATE ( MIN ( 'Date'[Date] ) )
RETURN
CALCULATE (
DISTINCTCOUNT ( MyFactTable[ProductID] ),
FILTER (
MyFactTable,
MyFactTable[StartDate] <= _startdate
&& MyFactTable[EndDate] >= _enddate
),
MyFactTable[IsActive] = 1
)
2.If you use as first parameter of FILTER an entire Fact Table, Storage Engine will load in memory the Expanded Table which is very expensive. Therefore, as a second step the formula should become:
_Calc_Count Fields 2:=
VAR _startdate = CALCULATE ( MAX ( 'Date'[Date] ) )
VAR _enddate = CALCULATE ( MIN ( 'Date'[Date] ) )
RETURN
CALCULATE (
DISTINCTCOUNT ( MyFactTable[ProductID] ),
MyFactTable[StartDate] <= _startdate && MyFactTable[EndDate] >= _enddate,
MyFactTable[IsActive] = 1
)
Next, based on the preliminary test you can decide where to invest your effort.
The issue is the DISTINCTCOUNT:
- explore some alternative algorithms for approximating DISTINCTCOUNT (HIGH EFFORT)
- try to sort in the data source (back-end) the table by ProductId to allow better compression in AAS
- make sure ProductId is a Integer Data type with Encoding Hint: Value
The issue is in the FILTER:
- Try to change the "&&" with "," (LOW EFFORT)
- Investigate the cardinality of StartDate and EndDate. If they are DateTime, remove the Time part. (LOW EFFORT)
- Try to change the datasource in the back-end and sort by useful fields (for example, StartDate asc, so when AAS will read the table might perform better compression (LOW EFFORT)
- Make sure StartDate and Date are Whole Number data types, with Encoding Hint: Value (LOW EFFORT)
Overview of the table in question
I need to get a distinct count of the column Fkey_Dim_Resource_ID that has holiday to spare.
My Table consists of five columns:
Resource_Allocated_Holiday_ID (Primary Key)
Fkey_Dim_Resource_ID
Fkey_Dim_HolidayYear_ID
Fkey_Dim_Company_ID
Allocated_Holiday_Hrs_Qty
Measure:
Allocated Holiday (Hrs):= Var X= SUM([Allocated_Holiday_Hrs_Qty])
Return if(X =0; BLANK();X)
This measure below then uses the above, and the holiday spent from another metric:
Remaining Holiday (Hrs):= Var X = 'HolidayEntry Numbers'[Allocated Holiday (Hrs)] - [#Holiday Hours]
Return if(X=0;BLANK();X)
And now, I would like a metric that gives me the distinct count of Fkey_Dim_ResourceID where 'Remaining Holiday (hrs)' >0.
I have tried a lot of different stuff, but cannot seem to get it right.
test:=
ADDCOLUMNS(
SUMMARIZE('HolidayEntry Numbers'
;'HolidayEntry Numbers'[Fkey_Dim_Company_ID]
;'HolidayEntry Numbers'[Fkey_Dim_Resource_ID];
'HolidayEntry Numbers'[Fkey_Dim_HolidayYear_Id]
)
;"RemainingHoliday"; sum( [Remaining Holiday (Hrs)])
)
I would like for a distinct count of Fkey_Dim_Resource_ID that has holiday left, that takes into account the context.
Thanks in advance.
With this measure:
test4 virker når ressourcen er med:=COUNTROWS (
FILTER (
ADDCOLUMNS (
VALUES ( 'HolidayEntry
Numbers'[Fkey_Dim_Resource_ID]);
"remholiday"; CALCULATE ( [Remaining Holiday
(Hrs)] )
);
[remholiday] > 0
)
)
I get the following result:
Result of the advice1
So the metric works, when in the context of a Resource, but not when in the context of a Fkey_dim_holiday_Year_ID.
Thanks ion advance.
Resources with remaining holiday hours =
COUNTROWS ( // counts rows in a table
FILTER ( // returns a table, filtering based on predicate
// below is unique values of the column in context, as a
// one-column table
VALUES ( 'HolidayEntry Numbers'[Fkey_Dim_Resource_ID] ),
[Remaining Holiday (hrs)] > 0 // keep rows meeting this criterion
)
)
As a matter of style, you should fully qualify column names as 'Table'[Column], and never fully qualify measure references, i.e. don't prefix with table name. This conforms with all style guides I know, and helps to ensure your code is unambiguous (since both columns and measures are referenced in square brackets).
I need to be able to sum availability based on product and say show me top 3, and categorize the rest as Others. I have two tables in a matrix connected by a product table.
I tried so many ways -
i was able to create this measure for July (which is what i will be sorting with) - I get the correct ranking column for July.
i know i'm missing something. i tried to take that ranking measure statement and add an if statement and couldn't get it to do the ranking.
the picture would make more sense *(my formulas are based on actual column names)
Partner Ranking =
VAR summry =
SUMMARIZE (
ALLSELECTED ( Latest ),
[partner_group],
"Sum", COUNT ( Latest[site_url] )
)
VAR tmp =
ADDCOLUMNS ( summry, "RNK", RANKX ( summry, [Sum],, DESC, DENSE ) )
RETURN
MAXX (
FILTER ( tmp, [partner_group] = SELECTEDVALUE ( Latest[partner_group] ) ),
[RNK]
)
I don't know what to do next. how can i do this when i have a separate table that is the product name that links the two tables?
I have a fact table with settlement_date, product_id, service_id, location_id, and ticket_id and srv_adjusted_earning columns.
I have determined the DAX query to generate a calculated column that sums the srv_adjusted_earning column over the date range: settlement date and settlement date - 27 days (i.e. a 4 week window) as:
=CALCULATE(
SUM(factService[SRV_ADJUSTED_EARNING]),
DATESBETWEEN
(
factService[SETTLEMENT_DATE],
DATEADD(factService[SETTLEMENT_DATE], -27, DAY),
factService[SETTLEMENT_DATE]
),
FILTER(factService, factService[PRO_ID] = EARLIER(factService[PRO_ID])),
FILTER(factService, factService[SER_ID] = EARLIER(factService[SER_ID])),
FILTER(factService, factService[LOC_ID_SELLING] =
EARLIER(factService[LOC_ID_SELLING])),
FILTER(factService, factService[TIS_ID] = EARLIER(factService[TIS_ID]))
)
I am trying to convert this DAX calculated column to a measure and I tried the following:
blob:=CALCULATE
(
SUM(factService[SRV_ADJUSTED_EARNING]),
DATESBETWEEN
(
factService[SETTLEMENT_DATE],
DATEADD(factService[SETTLEMENT_DATE], -27, DAY),
factService[SETTLEMENT_DATE]
),
ALLEXCEPT(factService, factService[PRO_ID]),
ALLEXCEPT(factService, factService[SER_ID]),
ALLEXCEPT(factService, factService[LOC_ID_SELLING]),
ALLEXCEPT(factService, factService[TIS_ID])
)
But I get:
Error: Calculation error in measure 'factService'[blob]: A single value for column 'SETTLEMENT_DATE' in table 'factService' cannot be determined. This can happen when a measure formula refers to a column that contains many values without specifying an aggregation such as min, max, count, or sum to get a single result.
Anybody know how I fix this?
As the error mentions, the issue is with factService[SETTLEMENT_DATE]. In the measure, there is no row context so that it knows which date you are talking about, so you need to specify it somehow. I'd suggest using a variable along these lines:
blob :=
VAR SettleDate = MAX ( factService[SETTLEMENT_DATE] )
RETURN
CALCULATE (
SUM ( factService[SRV_ADJUSTED_EARNING] ),
DATESBETWEEN (
factService[SETTLEMENT_DATE],
SettleDate - 27,
SettleDate
),
ALLEXCEPT (
factService,
factService[PRO_ID],
factService[SER_ID],
factService[LOC_ID_SELLING],
factService[TIS_ID]
)
)
Here the variable picks the maximal settlement date in the current filter context. If that's not exactly what you need, adjust the definition accordingly.
In a Tabular SSAS Model, I'm trying to count the number of distinct customers that purchased a given product wtihin a YTD Timeframe. The table contains measures that aren't explicit sums, so I get the Cartesian Product of all products for each customer, regardless of no sales. I'm attempting to limit the count by filtering out customer / product combinations with YTD Sales = 0. However, I cannot get the FILTER to recognize the DATESYTD context. It only ever filters based upon Sales existing within the chosen calendar month. I've tried inserting the ALL function every which way.
This is what I have so far.
Measure:
CALCULATE (
DISTINCTCOUNT ( Fact[Customer] ),
DATESYTD ( Calendar[Date] ),
FILTER ( Fact,
CALCULATE ( [Sum of Sales], DATESYTD ( Calendar[Date] ) ) <> 0
)
)
This measure will, for example, count distinct customers purchasing a product in Month #5 if Month #5 is explicitly chosen. It will not, however, include a customer that purchased that item in Month #2 of the same year.
I think the following DAX should do the trick:
COUNTROWS(
FILTER(
VALUES(Fact[Customer]),
CALCULATE ( [Sum of Sales], DATESYTD ( Calendar[Date] ) ) <> 0
)
)
Also, make sure your 'Calendar' table has been marked as a date table. If, for some reason, you prefer not to mark it as a date table, rewrite the above DAX to:
COUNTROWS(
FILTER(
VALUES(Fact[Customer]),
CALCULATE ( [Sum of Sales], DATESYTD ( Calendar[Date] ), ALL('Calendar') ) <> 0
)
)
Edit: Do you have records in your fact table where [Sum of Sales] is 0? If not, then you could simplify and improve the performance considerably by writing:
CALCULATE(
DISTINCTCOUNT(Fact[Customer]),
DATESYTD( Calendar[Date] )
)
Again, if you haven't marked your 'Calendar' table as a date table, add ALL(Calendar) to remove the filter on specific calendar columns.