Time aggregation: time_agg
Syntax
time_agg ( periodIndTo { , periodIndFrom } { , op } { , first | last } )
Input parameters
op |
the scalar value, the Component or the Data Set to be converted.
If not specified, then time_agg is used in combination within an aggregation operator
|
periodIndFrom |
the source period indicator |
periodIndTo |
the target period indicator |
Examples of valid syntaxes
sum ( DS group all time_agg ( “A” ) )
time_agg ( “A”, cast ( “2012Q1”, time_period , ”YYYY\Qq” ) )
time_agg(“M”, cast (“2012-12-23”, date, “YYYY-MM-DD”) )
time_agg(“M”, DS1)
ds_2 := ds_1[calc Me1 := time_agg(“M”,Me1)]
Semantics for scalar operations
The operator converts a time, date or time_period value from a smaller to a larger duration.
Input parameters type
op
dataset { identifier < time > _ , identifier _* }
| component<time>
| time
periodIndFrom, periodIndTo
duration
Result type
result
dataset { identifier < time > _ , identifier _* }
| component<time>
| time
Additional Constraints
If op is a Data Set then it has exactly one Identifier of type time, date or time_period and may have other Identifiers.
If time_agg is used in combination with an aggregation operator, op must not be specified, and the source dataset must have exactly one Identifier of type time, date or time_period (it may have additional Identifiers of other types).
It is only possible to convert smaller duration values to larger duration values (e.g. it is possible to convert monthly data to annual data but the contrary is not allowed).
Behaviour
The scalar version of this operator takes as input a time, date or time_period value, converts it to periodIndTo and returns a scalar of the corresponding type.
The Data Set version acts on a single Measure Data Set of type time, date or time_period and returns a Data Set having the same structure.
Finally, VTL also provides a component version, for use in combination with an aggregation operator, because the change of frequency requires an aggregation. In this case, the operator converts the period_indicator of the data points (e.g., convert monthly data to annual data).
On time type, the operator maps the input value into the comprising larger regular interval, whose duration is the one specified by the periodIndTo parameter.
On date type, the operator maps the input value into the comprising larger period, whose duration is the one specified by the periodIndTo parameter, which is conventionally represented either by the start or by the end date, according to the first/last parameter.
On time_period type, the operator maps the input value into the comprising larger time period specified by the periodIndTo parameter (the original period indicator is converted in the target one and the number of periods is adjusted correspondingly).
The input duration periodIndFrom is optional. In case of time_period Data Points, the input duration can be inferred from the internal representation of the value. In case of time or date types, it is inferred by the implementation.
Examples
Given the operand dataset DS_1:
Input DS_1 (see structure)
Id_1 |
Id_2 |
Me_1 |
|---|---|---|
2010Q1 |
A |
20 |
2010Q2 |
A |
20 |
2010Q3 |
A |
20 |
2010Q1 |
B |
50 |
2010Q2 |
B |
50 |
2010Q1 |
C |
10 |
2010Q2 |
C |
10 |
Example 1
DS_r := sum ( DS_1 group all time_agg ( "A" ));
results in (see structure):
Id_1 |
Id_2 |
Me_1 |
|---|---|---|
2010 |
A |
60 |
2010 |
B |
100 |
2010 |
C |
20 |
Example 2
DS_r := time_agg ( "Q", cast ( "2012M01", time_period, "YYYY\\MMM" ) );
returns “2012Q1”.
Example 3
The following example maps a date to quarter level, 2012 (end of the period):
DS_r := time_agg( “Q”, cast(“20120213”, date, “YYYYMMDD”), last );
and produces a date value corresponding to the string “20120331”.
Example 4
The following example maps a date to year level, 2012 (beginning of the period):
DS_r := time_agg(cast( “A”, “2012M1”, date, “YYYYMMDD”), first );
and produces a date value corresponding to the string “20120101”.