#802: AbstractAggregatedValues.java

Statistical Aggregation Base Class

A generic abstract base class for accumulating aggregated values over time. Used by the statistics module and specifically by LiquidityForecast for weighted-average payment timing calculations. Subclasses (IntAggregatedValues, BigDecimalAggregatedValues) provide type-specific implementations.

Generic parameter <T> allows arithmetic operations on any numeric type without code duplication. The class stores two parallel LinkedLists — one for values, one for weights — enabling both simple averages and weighted averages on the same data set.

Template Method Pattern

The class implements the Template Method pattern. Core algorithms (getAverage(), getWeightedAverage()) are defined in the abstract class, while type-specific arithmetic is delegated to subclasses via six abstract methods:

Dirty-Flag Caching

Both average and weightedAverage results are lazily computed and cached using a dirty-flag pattern. Two boolean fields (averageDirty, weightedAverageDirty) track whether the cached value is stale. Calling add() sets both dirty flags to true. The cached values are only recomputed when getAverage() or getWeightedAverage() is called and the respective dirty flag is set.

This optimization avoids recomputing averages on every addition, which is important when many values are added in sequence before the average is queried.

Fluent API (Method Chaining)

Both add() overloads and clear() return this, enabling fluent chaining:

aggregatedValues.add(val1).add(val2, weight2).add(val3).getWeightedAverage();

Usage Context: LiquidityForecast

In LiquidityForecast, this class is used to compute the weighted average payment time across multiple invoice positions. Each position contributes a value (amount) and a weight (days until payment). The weighted average yields a single representative payment time for the forecast.