package privacy.differentialprivacy.pipelinedp4j.proto

Get desktop application:
View/edit binary Protocol Buffers messages

A collection of accumulators for all metrics that can possibly be computed. Only accumulators that correspond to the aggregations, which are actually being computed, are populated.

• optional PrivacyIdCountAccumulator privacy_id_count_accumulator = 3

• optional CountAccumulator count_accumulator = 1

• optional SumAccumulator sum_accumulator = 2

• optional VectorSumAccumulator vector_sum_accumulator = 7

• optional MeanAccumulator mean_accumulator = 4

• optional QuantilesAccumulator quantiles_accumulator = 5

• optional VarianceAccumulator variance_accumulator = 6

Used in: CompoundAccumulator

• int64 count = 1

The anonymized aggregated result returned by PipelineDP4j. The proto has fields for all possible metrics that can be computed. Only metrics requested by the user are populated.

• double privacy_id_count = 3

• double count = 1

• double sum = 2

• repeated double vector_sum = 7

• double mean = 4

• repeated double quantiles = 5

  Sorted by rank.

• double variance = 6

Used in: CompoundAccumulator

• int64 count = 1

• double normalized_sum = 2

Contribution that belong to the same privacy ID and partition key.

• repeated PrivacyIdContributions.MultiValueContribution multi_value_contributions = 1

  Used for metrics like VECTOR_SUM where a single contribution has multiple double values.

• repeated double single_value_contributions = 2

  Used for metrics like SUM where a single contribution has a single double value.

Used in: PrivacyIdContributions

• repeated double values = 1

Used in: CompoundAccumulator

• int64 count = 1

Used in: CompoundAccumulator

• bytes serialized_quantiles_summary = 1

Used in: CompoundAccumulator

• double sum = 1

Used in: CompoundAccumulator

• int64 count = 1

• double normalized_sum = 2

• double normalized_sum_squares = 3

Used in: CompoundAccumulator

• repeated double sums_per_dimension = 1