Applies to:  CELONIS 4.7 

Description

LEAD returns the row that follows the current row by offset number of rows. It is possible to specify a column based ordering and partitioning. Null values are skipped.

Syntax

LEAD ( table.column [, ORDER BY ( order_column [sorting], ... )] [, PARTITION BY ( partition_column, ... )] [, offset ] )
  • column: The source column where following rows are taken from.
  • order_column: Optional order column to specify an sorting.
  • sorting: Each of these columns can have an optional tag specifying the ordering of the column. Default is ascending:
    • ASC: Ascending order
    • DESC: Descending order
  • partition_column: Optional partition column to specify groups in which LEAD should operate.
  • offset: The number of non-NULL rows following the current row. The default value is 1.

Ordering

One or more columns can be given to specify an ordering. This tells the LEAD function what the following element actually is. Optionally every column can be tagged as ascending or descending.


[1] LEAD with a single order column.

Query
Column1
LEAD ( "Table1"."column" , ORDER BY ( "Table1"."order" ) )
Input
Table1
column : STRINGorder : INT
'A'
3
'B'
2
'C'
5
'D'
1
'E'
4
Output
Result
Column1 : STRING
'E'
'A'
null
'B'
'C'



[2] LEAD with a single order column tagged as descending.

Query
Column1
LEAD ( "Table1"."column" , ORDER BY ( "Table1"."order" DESC ) )
Input
Table1
column : STRINGorder : INT
'A'
3
'B'
2
'C'
5
'D'
1
'E'
4
Output
Result
Column1 : STRING
'B'
'D'
'E'
null
'A'



[3] LEAD with multiple order columns.

Query
Column1
LEAD ( "Table1"."column" , ORDER BY ( "Table1"."order1" , "Table1"."order2" ) )
Input
Table1
column : STRINGorder1 : INTorder2 : STRING
'A'
3
'z'
'B'
1
'z'
'C'
1
'y'
'D'
2
'y'
'E'
2
'x'
Output
Result
Column1 : STRING
null
'E'
'B'
'A'
'D'


Partitioning

The partition columns specify groups. The LEAD function operates independently within every group. This means when an ordering is given it is applied within every group and the last offset elements in every group have a leading value of null.


[4] LEAD with a single partition column.

Query
Column1
LEAD ( "Table1"."column" , PARTITION BY ( "Table1"."Country" ) )
Input
Table1
column : INTCountry : STRING
1
'Germany'
2
'Germany'
3
'Germany'
1
'USA'
2
'USA'
Output
Result
Column1 : INT
2
3
null
2
null



[5] LEAD with multiple partition columns.

Query
Column1
LEAD ( "Table1"."column" , PARTITION BY ( "Table1"."Country" , "Table1"."State" ) )
Input
Table1
column : INTCountry : STRINGState : STRING
1
'Germany''Bavaria'
2
'Germany''Berlin'
2
'Germany''Bavaria'
3
'Germany''Bavaria'
1
'USA''California'
Output
Result
Column1 : INT
2
null
3
null
null



[6] LEAD with multiple partition columns.

Query
Column1
LEAD (
    "Table1"."column" ,
    ORDER BY ( "Table1"."year" ) ,
    PARTITION BY ( "Table1"."Country" , "Table1"."State" )
)
Input
Table1
column : INTyear : DATECountry : STRINGState : STRING
1
Tue Dec 31 2002 00:00:00.000
'Germany''Bavaria'
2
Fri Dec 31 1999 00:00:00.000
'Germany''Berlin'
2
Mon Dec 31 2001 00:00:00.000
'Germany''Bavaria'
3
Fri Dec 31 1999 00:00:00.000
'Germany''Bavaria'
1
Fri Dec 31 1999 00:00:00.000
'USA''California'
Output
Result
Column1 : INT
null
null
1
2
null


NULL handling

The leading value for a NULL value is the same value as the leading value of the last non-NULL value. The offset parameter counts only non-NULL values.


[7] Example for LEAD applied to a column including NULL values.

Query
Column1
LEAD ( "Table1"."column" , PARTITION BY ( "Table1"."Country" ) )
Input
Table1
column : INTCountry : STRING
1
'Germany'
null
'Germany'
3
'Germany'
4
'Germany'
1
'USA'
2
'USA'
null
'USA'
Output
Result
Column1 : INT
3
3
4
null
2
null
null



[8] Offset ignores NULL values.

Query
Column1
LEAD ( "Table1"."column" , PARTITION BY ( "Table1"."Country" ) , 2 )
Input
Table1
column : INTCountry : STRING
1
'Germany'
null
'Germany'
3
'Germany'
4
'Germany'
1
'USA'
2
'USA'
null
'USA'
null
'USA'
5
'USA'
Output
Result
Column1 : INT
4
4
null
null
5
null
null
null
null


Examples


[9] Simple example of LEAD returning the following row. If a row does not have a following row NULL is returned.

Query
Column1
LEAD ( "Table1"."column" )
Input
Table1
column : STRING
'C'
'D'
'A'
'E'
'B'
Output
Result
Column1 : STRING
'D'
'A'
'E'
'B'
null



[10] LEAD with an offset.

Query
Column1
LEAD ( "Table1"."column" , 3 )
Input
Table1
column : STRING
'C'
'D'
'A'
'E'
'B'
Output
Result
Column1 : STRING
'E'
'B'
null
null
null



[11] LEAD can be used to simulate ACTIVITY_LEAD.

Query
Column1
LEAD ( "Table1"."activity" , ORDER BY ( "Table1"."timestamp" ) , PARTITION BY ( "Table1"."case" ) )
Input
Table1
case : INTactivity : STRINGtimestamp : DATE
1
'A'
Mon Feb 01 2016 01:00:00.000
1
'B'
Mon Feb 01 2016 02:00:00.000
1
'C'
Mon Feb 01 2016 03:00:00.000
2
'A'
Mon Feb 01 2016 01:00:00.000
2
'B'
Mon Feb 01 2016 02:00:00.000
2
'C'
Mon Feb 01 2016 03:00:00.000
2
'D'
Mon Feb 01 2016 04:00:00.000
Output
Result
Column1 : STRING
'B'
'C'
null
'B'
'C'
'D'
null



[12] It is possible to combine an arbitrary amount of order columns with an arbitrary amount of partition columns.

Query
Column1
LEAD (
    "Table1"."column" ,
    ORDER BY ( "Table1"."year" DESC , "Table1"."value" ) ,
    PARTITION BY ( "Table1"."Country" , "Table1"."State" )
)
Input
Table1
column : INTyear : DATEvalue : FLOATCountry : STRINGState : STRING
1
Tue Dec 31 2002 00:00:00.000
1.0
'Germany''Bavaria'
2
Fri Dec 31 1999 00:00:00.000
2.0
'Germany''Berlin'
2
Mon Dec 31 2001 00:00:00.000
3.0
'Germany''Bavaria'
3
Fri Dec 31 1999 00:00:00.000
4.0
'Germany''Bavaria'
1
Fri Dec 31 1999 00:00:00.000
5.0
'USA''California'
1
Sun Dec 31 2000 00:00:00.000
6.0
'Germany''Berlin'
5
Tue Dec 31 2002 00:00:00.000
7.0
'USA''California'
6
Mon Dec 31 2001 00:00:00.000
8.0
'Germany''Berlin'
4
Tue Dec 31 2002 00:00:00.000
9.0
'Germany''Bavaria'
7
Wed Dec 31 2003 00:00:00.000
10.0
'USA''California'
Output
Result
Column1 : INT
4
null
3
null
null
2
1
1
2
5



[13] All optional parameters can be combined.

Query
Column1
LEAD (
    "Table1"."column" ,
    ORDER BY ( "Table1"."year" DESC , "Table1"."value" ) ,
    PARTITION BY ( "Table1"."Country" , "Table1"."State" ) ,
    2
)
Input
Table1
column : INTyear : DATEvalue : FLOATCountry : STRINGState : STRING
1
Tue Dec 31 2002 00:00:00.000
1.0
'Germany''Bavaria'
2
Fri Dec 31 1999 00:00:00.000
2.0
'Germany''Berlin'
2
Mon Dec 31 2001 00:00:00.000
3.0
'Germany''Bavaria'
3
Fri Dec 31 1999 00:00:00.000
4.0
'Germany''Bavaria'
1
Fri Dec 31 1999 00:00:00.000
5.0
'USA''California'
1
Sun Dec 31 2000 00:00:00.000
6.0
'Germany''Berlin'
5
Tue Dec 31 2002 00:00:00.000
7.0
'USA''California'
6
Mon Dec 31 2001 00:00:00.000
8.0
'Germany''Berlin'
4
Tue Dec 31 2002 00:00:00.000
9.0
'Germany''Bavaria'
7
Wed Dec 31 2003 00:00:00.000
10.0
'USA''California'
Output
Result
Column1 : INT
2
null
null
null
null
null
null
2
3
1


See also:

  • No labels