I am new to OLAP, I just designed my first cube and now I am trying to make some simple analyses. I have a simple time dimension that looks like this:
- Year
- Month
- DayOfWeek
Now I would like to sum all measures for e. g. Monday. I don't want to know the measure for Monday for every month of every year seperated (I am able to do this) but for all Mondays ever. Just one value.
Is it possible and if, how?
Thanks!
Time to break out the AdventureWorks sample and look at their time dimension. You need more attributes on your dimension. There should be attributes for Year, Month and Date that are arranged in a hierarchy, but the attribute for DayofWeek, MonthofYear, etc, should be separate if you need to aggregate on them.
Related
I am trying to create a calculation measure to compare revenue from one date to last year's equivalent date.
I have this "last year's equivalent date" stored in my date dimension.
So for example, my fact table of DailySummary has a measure of TotalRevenue and it is granular by date. If I wanted to show today's revenue (12/06/2019), the equivalent date from last year that is stored in the Date dimension would be 12/07/2018. I would want to show both those day's revenue side by side. I would then create other measures, like growth.
I am sure it is obvious, but if it helps, the DailySummary fact and the Date dimension are joined by the date, so today's date, not the last year equivalent.
Let me know if this doesn't make sense or if you need more information.
I think that you will need to use the PreviousPeriod MDX function.
In our date dimension, there are Month and Year attributes. Month attribute has set int month_year as key, to avoid duplicate problem, and Month attribute has name like January, February...
Now there is problem, if user drags Year and Month to filter, and would define one year, say 2017, he expect to see Month Attribute in filter only shows 12 month names in 2017. However, Month Attributes shows month names for all years, the names repeating many times.
There is already calendar hierarchy built in this dimension. But the user wants to define filter outside of hierarchy. How to build such Month Attribute in cube? Does some one have any idea about it? Thanks in advance!
The problem has been solved. Just add a new column in date table, say Month_in_current_Year, has the same content as Month column. Add this new column to date dimension as a new attribute. This new attribute works for my purpose. It shows only 12 elements as months in one year. And it slices correct value while setting it to filter. What magic is the cube!
I have a fact table with payroll data and contains columns such employee id, dollars, start date & end date (pay period). So the granularity is not at the daily level. How can we add this fact table to my cube and link to date dimension? I have a date a typical date dimension in the cube with date, month, quarter and year.
note - the start and end date do not always fall on the sames days of the calendar month. I know it is terrible idea to somehow "convert" the grain to daily level by diving the dollar amounts by the number of days between start date and end date but i can't figure out another/better option.
As explained in the documentation, a relationship of a fact table/measure group to a dimension need not point to the key attribute/column of the dimension, it may point to any attribute (like the month attribute in your case). Then, you only can use the attribute that you linked to and attributes which have direct or indirect references from it. If you would have a reference from month to quarter and from quarter to year in your dimension, then the measures from your measure group referencing the month attribute can use quarter or year. But they cannot use days.
I want to summarize sql query result week wise and month wise at the same time in a grid view. Is this even possible or I am just dreaming?
Requirement:
Show last one month data and next two month's data week wise in the grid.
Example-
If the current month is September then I want to show data from 1st August to 31st October categorized in weeks.
Show the data after the next month of current month in month wise view in the same grid.
Example-
data for month November and December will be shown categorized in month not in weeks.
grid or result should look something like below -
Please suggest something to achieve this
If you want a maintainable solution, use two independent queries, one for weekly aggregation other for monthly. Depending on the input run corresponding query.
I think you must work with group by or group by ... cube
SELECT x , y from <tabel> GROUP BY date_feld( to_char( 'MM' ))
but i don't now your DBMS so i can't give you a exact example for the date handling.
This is regarding calculating the sum for a set of values starting from the date a measure has entries till the current date.
So far what I could find was the YTD function. This limits the aggregation capability till the current date beginning from the first day of the year. But what my requirement is to start the calculation from the first value, this could be in the last year or may be two years before.
Eg:
Date-----------Value
11/9/2010-----2000
2/10/2011-----500
8/5/2011------1000
With YTD the value is: 1500
What I need is: 3500
I really appreciate any help on this.
Something like:
SUM([Date].[Day].AllMembers, [Measures].[Value])
OR
SUM(OpeningPeriod([Date].[Day]):ClosingPeriod([Date].[Day]), [Measures].[Value]),
where [Date].[Day] - it's a level DAY of your dimension;
[Measures].[Value] - your measure.