Recordwise Adjustment Factors

The template allows applying factor values defined in one ledger to the matching records in another ledger. The load factors input file should have the same structure as a regular ledger. However, factor records should have a distinct type that differs from other financial records (e.g. factor records can have a “Factor” type, while other financial records have “Loss”, “Premium”, and other types).

Suppose the load factors ledger has been created in S3:

s3://example_bucket/uploads/ledgers/example_ledger_upload/load_factors.parquet

The general structure of the Recordwise Adjustment Factors template in Graphene is:

{
    "_schema": "RecordwiseAdjustmentFactors_1.0",
    "factor_type_name": "Factor",
    "path": "s3://example_bucket/uploads/ledgers/example_ledger_upload/load_factors.parquet",
    "match_by": ["Event"]
}

Parameters

The parameters are defined as follows:

Parameter Name	Required	Type	Description
factor_type_name	Yes	string	Record type that will be used to identify factor records. All other record types will be considered as financial records.
path	Yes	string	The unescaped S3 key prefix or full S3 key that represents a complete ledger. This path may be relative to the bucket root or may be absolute.
match_by	Yes	string list	Ledger attribute name(s) that will be used to match financial records and factor records

Behavior

match_by currently list can be empty or can contain only one ledger attribute name, that should be used as an occurrence key.

In that case, financial records will be multiplied by factor records that have matching Trial, Time, and Occurrence Key. More than one attribute is not supported. If the list is empty, then all factor records are considered to have no occurrence key. In that case financial and factor records with empty occurrence keys and matching Trial and Time will be multiplied.

Files with factor records should be uploaded to the regular ledgers folder. Multi-partition files are supported like regular ledgers.

All values of the matching financial records are multiplied by the factor values in the input ledger. The result is the list of financial records in the output ledger. Factor records are not propagated to the output ledger. Currently, it supports matching financial records with factor records by Trial, Time, and Occurrence Key attributes only. Financial records that don’t have matching factor(s) are not emitted to the output ledger. For a single financial record that matches N multiple factors, N financial records will be generated. For M financial records and N matching factor records, N x M output financial records will be emitted.

Example

User uploads ledger 1 with losses and ledger 2 with factors.

Ledger 1:

Trial	Time	Type	Value	EventId
1	1	Loss	100	4
1	1	Loss	200	4
1	2	Loss	300	4

Ledger 2:

Trial	Time	Type	Value	EventId
1	1	Factor	1.1	4
1	1	Factor	2	4
1	3	Factor	3	4

Output ledger:

Trial	Time	Type	Value	EventId
1	1	Loss	110	4
1	1	Loss	220	4
1	1	Loss	200	4
1	1	Loss	400	4

Example notes:

• The third event from ledger 1 with a value of 300 does not have matching factors in ledger 2 and is therefore not included in the output ledger.

• The third factor with a value of 3 in ledger 2 does not apply to any values in ledger 1 because it does not match any loss records.

• Factors with values of 1.1 and 2 match two records in ledger 1, resulting in four output records: two losses multiplied by two factors.