MDX join with the same dimension - join

I'm writing some MDX to join a dimension to itself based on two different periods to get a common list, then do a count against this list for both.
In short, I need to
get a list of Student.UniqueId's for Period1 which has a flag (IsValid) that is set that isn't set within the Period2 data
get a full list of Students for Period2
join the two lists and produce two records (one for each period) with the same count (these counts will be used for calculated member calculations within each period)
I have tried doing it via subselect and exists clause with filter
SELECT
{
[Measures].[FactStudentCount]
} on COLUMNS,
{ NONEMPTY
(
[TestEvent].[TestEvents].[Name].ALLMEMBERS
* [TestEvent].[PeriodName].[PeriodName].ALLMEMBERS
)
} ON ROWS
FROM ( SELECT ( {
exists
(
filter([Student].[UniqueId].[UniqueId].MEMBERS
,([TestEvent].[Key].&[Period1], [IsValid].[Code].&[Yes]))
,
filter([Student].[UniqueId].[UniqueId].MEMBERS
,[TestEvent].[Key].&[Period2])
)
}) ON COLUMNS
FROM [MyCube])
...however this doesn't give the correct result
(To obtain context) I have also tried similar exists/filter within a where clause
SELECT
{
[Measures].[FactStudentCount]
} on COLUMNS,
{ NONEMPTY
(
[TestEvent].[TestEvents].[Name].ALLMEMBERS
* [TestEvent].[PeriodName].[PeriodName].ALLMEMBERS
)
} ON ROWS
FROM [MyCube]
where (
exists
(
filter([Student].[UniqueId].[UniqueId].MEMBERS
,([TestEvent].[Key].&[Period1], [IsValid].[Code].&[Yes]))
,
filter([Student].[UniqueId].[UniqueId].MEMBERS
,[TestEvent].[Key].&[Period2])
)
)
...however again this doesn't produce the correct result
I have tried tweaking the filter statements (within the exists) to something like
(filter(existing([Student].[UniqueId].[UniqueId].allmembers),[TestEvent].[Key].CurrentMember.MemberValue = 'Period1'), [IsValid].[Code].&[Yes])
,
(filter(existing([Student].[UniqueId].[UniqueId].allmembers),[TestEvent].[Key].CurrentMember.MemberValue = 'Period2'))
...however this only returns one row (for Period1) - that said it is the correct total
I have also tried via a CrossJoin with NonEmpty however it fails because the fields come from the same hierarchy - the message "The Key hierarchy is used more than once in the Crossjoin function"
Does any one have any insight into how to resolve the above scenario ?

This is what I did
NonEmpty(
NonEmpty(
{([Student].[UniqueId].[UniqueId].members)},{([TestEvent].[Key].&[Period1], [IsValid].[Code].&[Yes])}
)
,
{([Student].[UniqueId].[UniqueId].members,[TestEvent].[Key].&[Period2])}
)
This gets all Period1 elements, with IsValid='Yes' then 'left joins' this with records in Period2

Related

Populating Fact Tables(Data Warehouse) and Querying

I am not sure how to query my fact tables(covid and vaccinations), I populated the dimensions with dummy data, I am supposed to leave the fact tables empty? As far as I know, they would get populated when I write the queries.
I am not sure how to query the tables I have tried different things, but I get an empty result.
Below is a link to the schema.
I want to find out the "TotalDeathsUK"(fact table COVID) for the last year caused by each "Strain"(my strain table has 3 strain in total.
You can use MERGE to poulate your fact table COVIDFact :
MERGE
INTO factcovid
using (
SELECT centerid,
dateid,
patientid,
strainid
FROM yourstagingfacttable ) AS f
ON factcovid.centerid = f.centerid AND factcovid.dateid=f.dateid... //the join columns
WHEN matched THEN
do nothing WHEN NOT matched THEN
INSERT VALUES
(
f.centerid,
f.dateid,
f.patientid,
f.strainid
)
And for VaccinationsFact :
MERGE
INTO vaccinations
using (
SELECT centerid,
dateid,
patientid,
vaccineid
FROM yourstagingfacttable ) AS f
ON factcovid.centerid = f.centerid //join condition(s)
WHEN matched THEN
do nothing WHEN NOT matched THEN
INSERT VALUES
(
f.centerid,
f.dateid,
f.patientid,
f.vaccineid
)
For the TotalDeathUK measure :
SELECT S.[Name] AS Strain, COUNT(CF.PatientID) AS [Count of Deaths] FROM CovidFact AS CF
LEFT JOIN Strain AS S ON S.StrainID=CF.StrainID
LEFT JOIN Time AS T ON CF.DateID=T.DateID
LEFT JOIN TreatmentCenter AS TR ON TR.CenterID=CF.CenterID
LEFT JOIN City AS C ON C.CityID = TR.CityID
WHERE C.Country LIKE 'UK' AND T.Year=2020
AND Result LIKE 'Death' // you should add a Result column to check if the Patient survived or died
GROUP BY S.[Name]

How to join two completely different cubes in MDX?

I want to join two completely different cubes in mdx (I am using MS SSRS 2008). I am really new to mdx and I have no idea how to do it. I want to join by SKU if possible. Can any body please tell me how to do it?
mdx Query 1
SELECT NON EMPTY { [Measures].[Sales], [Measures].[Quantity] } ON COLUMNS,
NON EMPTY { ([Date YMD].[Day].ALLMEMBERS *
[Regions And Stores].[Store Name].[Store Name].ALLMEMBERS *
[Products].[Products].ALLMEMBERS *
[SKU].[SKU].ALLMEMBERS ) } DIMENSION PROPERTIES MEMBER_CAPTION, MEMBER_UNIQUE_NAME ON ROWS
FROM [Super] CELL PROPERTIES VALUE, BACK_COLOR, FORE_COLOR, FORMATTED_VALUE, FORMAT_STRING, FONT_NAME, FONT_SIZE, FONT_FLAGS
mdx Query 2
SELECT NON EMPTY { [Measures].[Quantity] } ON COLUMNS,
NON EMPTY { ([Store Name].[Store Name].ALLMEMBERS *
[Products].[Products].ALLMEMBERS *
[SKU].[SKU].ALLMEMBERS ) } DIMENSION PROPERTIES MEMBER_CAPTION, MEMBER_UNIQUE_NAME ON ROWS
FROM [Inventory Activity] CELL PROPERTIES VALUE, BACK_COLOR, FORE_COLOR, FORMATTED_VALUE, FORMAT_STRING, FONT_NAME, FONT_SIZE, FONT_FLAGS
Any help will highly appreciated.
Thank you
In SSRS you need to load two Datasets in your report and join them in the tablix. For example:
Load Dataset1 and Dataset2 into your report, with the column ID which links Dataset1 to Dataset2. Then put a tablix in your report. Display Dataset1 in your tablix. Now add a new column to your tablix and add the following expression:
=Lookup(Fields!Dataset1ID.Value, Fields!Dataset2ID.Value, Fields!SalesAmount.Value, "Dataset2")
The expression works as follows:
- First argument is the foreign key column from Dataset1
Second argument is corrspondending key column from Dataset2
Third argument is the column you want to display in the tablix which is from Dataset2
Forth argument is the name from the dataset you want to join with (Dataset2)
Here is the reference for the Lookup() function: https://learn.microsoft.com/de-de/sql/reporting-services/report-design/report-builder-functions-lookup-function?view=sql-server-2017
Welcome to MDX, I guess you are looking for SQL Join equivalent in MDX. However MDX doesnt support joins like SQL. One way to solve the issue is to retrieve the data via ADOMD into data cells and then join them in memory. However I would like to know the scenario, which requires you to join results from two cubes.

How to convert SQL statement "delete from TABLE where someID not in (select someID from Table group by property1, property2)

I'm trying to convert the following SQL statement to Core Data:
delete from SomeTable
where someID not in (
select someID
from SomeTable
group by property1, property2, property3
)
Basically, I want to retrieve and delete possible duplicates in a table where a record is deemed a duplicate if property1, property2 and property3 are equal to another record.
How can I do that?
PS: As the title says, I'm trying to convert the above SQL statement into iOS Core Data methods, not trying to improve, correct or comment on the above SQL, that is beyond the point.
Thank you.
It sounds like you are asking for SQL to accomplish your objective. Your starting query won't do what you describe, and most databases wouldn't accept it at all on account of the aggregate subquery attempting to select a column that is not a function of the groups.
UPDATE
I had initially thought the request was to delete all members of each group containing dupes, and wrote code accordingly. Having reinterpreted the original SQL as MySQL would do, it seems the objective is to retain exactly one element for each combination of (property1, property2, property3). I guess that makes more sense anyway. Here is a standard way to do that:
delete from SomeTable st1
where someID not in (
select min(st2.someId)
from SomeTable st2
group by property1, property2, property3
)
That's distinguished from the original by use of the min() aggregate function to choose a specific one of the someId values to retain from each group. This should work, too:
delete from SomeTable st1
where someID in (
select st3.someId
from SomeTable st2
join SomeTable st3
on st2.property1 = st3.property1
and st2.property2 = st3.property2
and st2.property3 = st3.property3
where st2.someId < st3.someId
)
These two queries will retain the same rows. I like the second better, even though it's longer, because the NOT IN operator is kinda nasty for choosing a small number of elements from a large set. If you anticipate having enough rows to be concerned about scaling, though, then you should try both, and perhaps look into optimizations (for example, an index on (property1, property2, property3)) and other alternatives.
As for writing it in terms of Core Data calls, however, I don't think you exactly can. Core Data does support grouping, so you could write Core Data calls that perform the subquery in the first alternative and return you the entity objects or their IDs, grouped as described. You could then iterate over the groups, skip the first element of each, and call Core Data deletion methods for all the rest. The details are out of scope for the SO format.
I have to say, though, that doing such a job in Core Data is going to be far more costly than doing it directly in the database, both in time and in required memory. Doing it directly in the database is not friendly to an ORM framework such as Core Data, however. This sort of thing is one of the tradeoffs you've chosen by going with an ORM framework.
I'd recommend that you try to avoid the need to do this at all. Define a unique index on SomeTable(property1, property2, property3) and do whatever you need to do to avoid trying to creating duplicates or to gracefully recover from a (failed) attempt to do so.
DELETE SomeTable
FROM SomeTable
LEFT OUTER JOIN (
SELECT MIN(RowId) as RowId, property1, property2, property3
FROM SomeTable
GROUP BY property1, property2, property3
) as KeepRows ON
SomeTable.RowId = KeepRows.RowId
WHERE
KeepRows.RowId IS NULL
A few pointers for doing this in iOS: Before iOS 9 the only way to delete objects is individually, ie you will need to iterate through an array of duplicates and delete each one. (If you are targeting iOS9, there is a new NSBatchDeleteRequest which will help delete them all in one go - it does act directly on the store but also does some cleanup to eg. ensure relationships are updated where necessary).
The other problem is identifying the duplicates. You can configure a fetch to group its results (see the propertiesToGroupBy of NSFetchRequest), but you will have to specify NSDictionaryResultType (so the results are NOT the objects themselves, just the values from the relevant properties.) Furthermore, CoreData will not let you fetch properties (other than aggregates) that are not specified in the GROUP BY. So the suggestion (in the other answer) to use min(someId) will be necessary. (To fetch an expression such as this, you will need to use an NSExpression, embed it in an NSExpressionDescription and pass the latter in propertiesToFetch of the fetch request).
The end result will be an array of dictionaries, each holding the someId value of your prime records (ie the ones you don't want to delete), from which you have then got to work out the duplicates. There are various ways, but none will be very efficient.
So as the other answer says, duplicates are better avoided in the first place. On that front, note that iOS 9 allows you to specify attributes that you would like to be unique (individually or collectively).
Let me know if you would like me to elaborate on any of the above.
Group-wise Maximum:
select t1.someId
from SomeTable t1
left outer join SomeTable t2
on t1.property1 = t2.property1
and t1.property2 = t2.property2
and t1.property3 = t2.property3
and t1.someId < t2.someId
where t2.someId is null;
So, this could be the answer
delete SomeTable
where someId not in
(select t1.someId
from SomeTable t1
left outer join SomeTable t2
on t1.property1 = t2.property1
and t1.property2 = t2.property2
and t1.property3 = t2.property3
and t1.someId < t2.someId
where t2.someId is null);
Sqlfiddle demo
You can use exists function to check for each row if there is another row that exists whose id is not equal to the current row and all other properties that define the duplicate criteria of each row are equal to all the properties of the current row.
delete from something
where
id in (SELECT
sm.id
FROM
sometable sm
where
exists( select
1
from
sometable sm2
where
sm.prop1 = sm2.prop1
and sm.prop2 = sm2.prop2
and sm.prop3 = sm2.prop3
and sm.id != sm2.id)
);
I think you could easily handle this by creating a derived duplicate_flg column and set it to 1 when all three property values are equal. Once that is done, you could just delete those records where duplicate_flg = 1. Here is a sample query on how to do this:
--retrieve all records that has same property values (property1,property2 and property3)
SELECT *
FROM (
SELECT someid
,property1
,property2
,property3
,CASE
WHEN property1 = property2
AND property1 = property3
THEN 1
ELSE 0
END AS duplicate_flg
FROM SomeTable
) q1
WHERE q1.duplicate_flg = 1;
Here is a sample delete statement:
DELETE
FROM something
WHERE someid IN (
SELECT someid
FROM (
SELECT someid
,property1
,property2
,property3
,CASE
WHEN property1 = property2
AND property1 = property3
THEN 1
ELSE 0
END AS duplicate_flg
FROM SomeTable
) q1
WHERE q1.duplicate_flg = 1
);
Simply, if you want to remove duplicate from table you can execute below Query :
delete from SomeTable
where rowid not in (
select max(rowid)
from SomeTable
group by property1, property2, property3
)
if you want to delete all duplicate records try the below code
WITH tblTemp as
(
SELECT ROW_NUMBER() Over(PARTITION BY Property1,Property2,Property3 ORDER BY Property1) As RowNumber,* FROM Table_1
)
DELETE FROM tblTemp where RowNumber >1
Hope it helps
Use the below query to delete the duplicate data from that table
delete from SomeTable where someID not in
(select Min(someID) from SomeTable
group by property1+property2+property3)

How to get row Count of the sqlite3_stmt *statement? [duplicate]

I want to get the number of selected rows as well as the selected data. At the present I have to use two sql statements:
one is
select * from XXX where XXX;
the other is
select count(*) from XXX where XXX;
Can it be realised with a single sql string?
I've checked the source code of sqlite3, and I found the function of sqlite3_changes(). But the function is only useful when the database is changed (after insert, delete or update).
Can anyone help me with this problem? Thank you very much!
SQL can't mix single-row (counting) and multi-row results (selecting data from your tables). This is a common problem with returning huge amounts of data. Here are some tips how to handle this:
Read the first N rows and tell the user "more than N rows available". Not very precise but often good enough. If you keep the cursor open, you can fetch more data when the user hits the bottom of the view (Google Reader does this)
Instead of selecting the data directly, first copy it into a temporary table. The INSERT statement will return the number of rows copied. Later, you can use the data in the temporary table to display the data. You can add a "row number" to this temporary table to make paging more simple.
Fetch the data in a background thread. This allows the user to use your application while the data grid or table fills with more data.
try this way
select (select count() from XXX) as count, *
from XXX;
select (select COUNT(0)
from xxx t1
where t1.b <= t2.b
) as 'Row Number', b from xxx t2 ORDER BY b;
just try this.
You could combine them into a single statement:
select count(*), * from XXX where XXX
or
select count(*) as MYCOUNT, * from XXX where XXX
To get the number of unique titles, you need to pass the DISTINCT clause to the COUNT function as the following statement:
SELECT
COUNT(DISTINCT column_name)
FROM
'table_name';
Source: http://www.sqlitetutorial.net/sqlite-count-function/
For those who are still looking for another method, the more elegant one I found to get the total of row was to use a CTE.
this ensure that the count is only calculated once :
WITH cnt(total) as (SELECT COUNT(*) from xxx) select * from xxx,cnt
the only drawback is if a WHERE clause is needed, it should be applied in both main query and CTE query.
In the first comment, Alttag said that there is no issue to run 2 queries. I don't agree with that unless both are part of a unique transaction. If not, the source table can be altered between the 2 queries by any INSERT or DELETE from another thread/process. In such case, the count value might be wrong.
Once you already have the select * from XXX results, you can just find the array length in your program right?
If you use sqlite3_get_table instead of prepare/step/finalize you will get all the results at once in an array ("result table"), including the numbers and names of columns, and the number of rows. Then you should free the result with sqlite3_free_table
int rows_count = 0;
while (sqlite3_step(stmt) == SQLITE_ROW)
{
rows_count++;
}
// The rows_count is available for use
sqlite3_reset(stmt); // reset the stmt for use it again
while (sqlite3_step(stmt) == SQLITE_ROW)
{
// your code in the query result
}

Grails GORM to return random rows from table?

In my grails application I have:
keywords = Keyword
.findAll("from Keyword where locale = '$locale' order by rand() ", [max:20])
Assume there are thousands of rows in the table that match the above criteria. But it seems the rows that are returned from the table are not random but in the order the rows are stored in Db although within the context of 20 rows that are returned they are random. For my application to work I want this query to return completely random rows from the table like it could be row id 203 , row id 3789, row id 9087, row id 789, and so on. How is that possible?
I use the following style:
Keyword.executeQuery('from Keyword order by rand()', [max: 9])
and it returns random rows from the entire table (we're using MySQL).
I'm not sure why execute query would behave differently from findAll though.
If you want to use a .withCriteria you can do that workaround:
User.withCriteria {
eq 'name', 'joseph'
sqlRestriction " order by rand()"
}
It's important to say that sometimes ( depends on the Criteria query created ) it's necessary to add a 1=1 in sqlRestriction because it adds an "and" condition in generated query.
So if you have a sqle exception use:
sqlRestriction " 1=1 order by rand()"

Resources