Applies to:  CELONIS 4.0 CELONIS 4.2 CELONIS 4.3 CELONIS 4.4 CELONIS 4.5 CELONIS 4.6 CELONIS 4.7 

Description

Filters can be defined as Analysis filters, Sheet filters or Component filters. If a query is sent to Celonis, all active filters are propagated to the requested table(s). Multiple filters on a table are merged together by a logical AND.

Filter Propagation

Filter propagation is necessary if there are one or more tables on which a filter is applied, which are not the same as the result table. In that case Celonis propagates the filters to the result table, along the specified join graph. For more information on the join graph see Joins).

Stable Filter

All filters in Celonis are stable. Stable in this context means that filters don't interfere with each other. As a result of this, the order of the filters doesn't matter. For example:

FILTER table.col > 400; FILTER table.col < 600;

returns the same as

FILTER table.col < 600; FILTER table.col > 400;

Filter stability is also the reason why Celonis doesn't support filters on aggregations, because they would not be stable. For example:

FILTER table.col > 400; FILTER AVG(table.col) < 400;

can return a different result than

FILTER AVG(table.col) < 400; FILTER table.col > 400;

Syntax

FILTER [FORCED] condition;

NULL handling

Applying a filter which compares a column against NULL or non-NULL always returns an empty result. To filter on all non-NULL values use the ISNULL function. NULL represents an unknown value. Celonis can not be sure if two unknown values are not the same.

Forced Filter

If a regular filter is set as a sheet or component filter, the affected Dropdown and Button Dropdown components still show all available values. If only those values which respect the filter should be displayed and selectable, a forced filter can be used. Analysis filters are forced by default.

Example

Dropdown component using this regular sheet or component filter:

FILTER "Table"."Country" IN ('DE','US');

All values are available in the dropdown menu. Values which do not match the filter condition are displayed in gray color.

Dropdown component using this forced sheet or component filter:

FILTER FORCED "Table"."Country" IN ('DE','US');

Only values which match the filter condition are available.

Examples


[1] Example where one filter is applied to the query. The filter condition excludes the second input row.

Query
Filter
FILTER "Numbers"."number" != 22;
Column1
"Numbers"."id"
Column2
"Numbers"."number"
Input
Numbers
id : INTnumber : INT
1
13
2
22
3
34
Output
Result
Column1 : INTColumn2 : INT
1
13
3
34



[2] Example where one filter is applied to the query. The filter condition excludes the second input row.

Query
Filter
FILTER "Numbers"."number" IN ( 13 , 34 );
Column1
"Numbers"."id"
Column2
"Numbers"."number"
Input
Numbers
id : INTnumber : INT
1
13
2
22
3
34
Output
Result
Column1 : INTColumn2 : INT
1
13
3
34



[3] Example where two filters are applied to the query. Both filter conditions are merged together by a logical AND. The first filter condition excludes the second input row, and the second filter condition excludes the first input row. Therefore, only the third row appears in the result.

Query
Filter
FILTER "Numbers"."number" IN ( 13 , 34 );
Filter
FILTER "Numbers"."id" IN ( 2 , 3 );
Column1
"Numbers"."id"
Column2
"Numbers"."number"
Input
Numbers
id : INTnumber : INT
1
13
2
22
3
34
Output
Result
Column1 : INTColumn2 : INT
3
34



[4] Example where one filter is applied to the query. The SUM aggregate function is applied after the filter has been applied to the input table.

Query
Filter
FILTER "Numbers"."number" IN ( 13 , 34 );
Column1
SUM ( "Numbers"."number" )
Input
Numbers
id : INTnumber : INT
1
13
2
22
3
34
Output
Result
Column1 : INT
47



[5] Example where two filters are applied to the query. Both filter conditions are merged together by a logical AND. The first filter condition excludes the first and third input row, and the second filter condition excludes the second and third input row. Therefore, the result is empty.

Query
Filter
FILTER "Numbers"."number" NOT IN ( 13 , 34 );
Filter
FILTER "Numbers"."id" = 1;
Column1
"Numbers"."id"
Column2
"Numbers"."number"
Input
Numbers
id : INTnumber : INT
1
13
2
22
3
34
Output
Result
Column1 : INTColumn2 : INT



[6] Example of two joined tables where one filter is applied to the query. The filter condition excludes the last row of the companyDetail input table, therefore, the last two rows of the caseTable are excluded.

Query
Filter
FILTER "companyDetail"."country" = 'DE';
Column1
"caseTable"."caseId"
Input
caseTable
caseId : INTcompanyCode : STRINGvalue : INT
1
'001'
600
2
'001'
400
3
'001'
200
4
'002'
300
5
'003'
300
6
'003'
200
companyDetail
companyCode : STRINGcountry : STRING
'001''DE'
'002''DE'
'003''US'

Foreign Keys
caseTable.companyCodecompanyDetail.companyCode
Output
Result
Column1 : INT
1
2
3
4



[7] Example of two joined tables where one filter is applied to the query. The filter condition excludes the last four rows of the caseTable.

Query
Filter
FILTER "caseTable"."value" > 300;
Column1
"caseTable"."caseId"
Column2
"companyDetail"."country"
Input
caseTable
caseId : INTcompanyCode : STRINGvalue : INT
1
'001'
600
2
'001'
400
3
'001'
200
4
'002'
300
5
'003'
300
6
'003'
200
companyDetail
companyCode : STRINGcountry : STRING
'001''DE'
'002''DE'
'003''US'

Foreign Keys
caseTable.companyCodecompanyDetail.companyCode
Output
Result
Column1 : INTColumn2 : STRING
1
'DE'
2
'DE'



[8] Example where DATE columns are being compared to DATE constants:

Query
Filter
FILTER "Astronomer"."Birth" < {d '1600-01-01' } AND "Astronomer"."Death" > {d '1599-12-31' };
Column1
"Astronomer"."Name"
Input
Astronomer
Name : STRINGBirth : DATEDeath : DATE
'Tycho Brahe'
Tue Dec 14 1546 00:00:00.000
Wed Oct 24 1601 00:00:00.000
'Giovanni Domenico Cassini'
Sun Jun 08 1625 00:00:00.000
Wed Sep 14 1712 00:00:00.000
'Galileo Galilei'
Tue Feb 15 1564 00:00:00.000
Wed Jan 08 1642 00:00:00.000
'Christiaan Huygens'
Sat Apr 14 1629 00:00:00.000
Fri Jul 08 1695 00:00:00.000
'Johannes Kepler'
Thu Dec 27 1571 00:00:00.000
Fri Nov 15 1630 00:00:00.000
Output
Result
Column1 : STRING
'Tycho Brahe'
'Galileo Galilei'
'Johannes Kepler'



[9] Example where we filter for DATE values from the last two weeks - and implicitly including possible timestamps in the future - by using a combination of ADD_DAYS and TODAY:

Query
Filter
FILTER "Table1"."Date" > ADD_DAYS ( TODAY ( ) , - 14 );
Column1
"Table1"."Value"
Input
Table1
Date : DATEValue : STRING
Sat Oct 31 2020 09:55:11.631
'20 days ago'
Thu Nov 05 2020 09:55:11.631
'15 days ago'
Tue Nov 10 2020 09:55:11.631
'10 days ago'
Sun Nov 15 2020 09:55:11.631
'5 days ago'
Fri Nov 20 2020 09:55:11.631
'Today'
Output
Result
Column1 : STRING
'10 days ago'
'5 days ago'
'Today'



[10] Applying a filter which compares a column against null always returns an empty result.

Query
Filter
FILTER Table1.Column1 = NULL;
Column1
Table1.Column1
Input
Table1
Column1 : INT
1
null
Output
Result
Column1 : INT

Comparison with NULL always returns NULL. To check for NULL values, please use <value> IS NULL or ISNULL(<value>)=1.



[11] Applying a filter which compares a column against not null always returns an empty result.

Query
Filter
FILTER Table1.Column1 != NULL;
Column1
Table1.Column1
Input
Table1
Column1 : INT
1
null
Output
Result
Column1 : INT

Comparison with NULL always returns NULL. To check for NULL values, please use <value> IS NULL or ISNULL(<value>)=1.


  • No labels