atoti.scope package

Submodules

atoti.scope.scope module

Scope class.

class atoti.scope.scope.Scope

Bases: abc.ABC

Specify a custom aggregation scope.

Module contents

Scope functions.

atoti.scope.cumulative(level, *, dense=False, partitioning=None, window=None)

Create a scope to be used in the computation of cumulative aggregations.

Cumulative aggregations include cumulative sums (also called running sum or prefix sum), mean, min, max, etc.

Example:

m2 = atoti.agg.sum(m1, scope=atoti.scope.cumulative(date))

date

m1

m2

2000/01/01

15

15

2000/01/02

10

25

2000/02/03

20

45

2000/02/05

30

75

2000/04/05

5

90

2000/04/05

10

95

If the level is part of a multi-level hierarchy, it is possible to reset the aggregation when the value of a level changes. For instance, a running sum over a date can be reset at the beginning of each month:

m2 = atoti.agg.sum(m1, scope=atoti.scope.cumulative("day", partitioning="month"))

year/month/day

m1

m2

Comment

2000/01/01

15

15

2000/01/02

10

25

2000/02/03

20

20

Reset at the beginning of February

2000/02/05

30

50

2000/04/05

5

5

Reset at the beginning of April

2000/04/05

10

15

Parameters
  • level (Level) – The level along which the aggregation is performed.

  • dense (bool) – When set to True, all members of the level, even those with no value for the

  • measure, will be taken into account for the cumulative aggregation (resulting (underlying) –

  • repeating values) (in) –

  • partitioning (Optional[Level]) – The levels in the hierarchy at which to start the aggregation over.

  • window (Union[range, Tuple[Optional[str], Optional[str]], None]) –

    The custom aggregation window. The window defines the set of members before and after a given member (using the level comparator) to be considered in the computation of the cumulative aggregation.

    The window can be a:

    • range starting with a <=0 value and ending with a >=0 value.

      By default the window is range(-∞, 0), meaning that the value for a given member is computed using all of the members before it and none after it.

    • time period as a two-element tuple starting with an offset of the form -xxDxxWxxMxxQxxY or None and ending with an offset of the form xxDxxWxxMxxQxxY or None.

      For instance, to compute the 5 previous days sliding mean:

      m2 = atoti.agg.mean(m1, scope=tt.scope.cumulative("date", window=("-5D", None)))
      

Return type

Scope

atoti.scope.origin(*levels)

Create an aggregation scope with an arbitrary number of levels.

The passed levels define a boundary above and under which the aggregation is performed differently. When those levels are not expressed in a query, the measure will drill down until finding the value for all members of these levels, and then aggregate those values using the user-defined aggregation function. This allows to compute measures that show the yearly mean when looking at the grand total, but the sum of each month’s value when looking at each year individually.

Parameters

levels (Level) – The levels defining the dynamic aggregation domain.

Return type

Scope

atoti.scope.siblings(hierarchy, *, exclude_self=False)

Create a “siblings” aggregation scope.

In a siblings scope, the value for the member of a given level in the hierarchy is computed by taking the contribution of all of the members on the same level (its siblings).

A siblings aggregation is an appropriate tool for operations such as marginal aggregations (marginal VaR, marginal mean) for non-linear aggregation functions.

Parameters
  • hierarchy (Hierarchy) – The hierarchy containing the levels along which the aggregation is performed.

  • exclude_self (bool) – Whether to include the current member’s contribution in its cumulative value.

Return type

Scope