Trifle.Stats
v2.5.0

View Source Trifle.Stats.Aggregator.Behaviour behaviour (Trifle.Stats v2.5.0)

Behaviour for custom aggregator implementations.

Aggregators process time-series data to compute aggregate values across data points. They take a series with timeline and values, and return aggregated results.

Example Implementation 

defmodule MyCustomAggregator do
  @behaviour Trifle.Stats.Aggregator.Behaviour

  @impl true
  def aggregate(series, path, slices \ 1) do
    # Extract values at the specified path
    values = series[:values] || []

    # Perform custom aggregation logic
    result = values
    |> Enum.map(&get_in(&1, String.split(path, ".")))
    |> Enum.reject(&is_nil/1)
    |> custom_aggregation_logic()

    # Return result in expected format
    [result]
  end

  defp custom_aggregation_logic(values) do
    # Your custom logic here
    Enum.sum(values) / length(values)
  end
end

Summary

Callbacks

aggregate(series, path)

Aggregate data from a series at the specified path (2-arity version).

aggregate(series, path, slices)

Aggregate data from a series at the specified path.

Callbacks

Link to this callback

aggregate(series, path)

View Source (optional) 
@callback aggregate(series :: map(), path :: String.t()) :: list()

Aggregate data from a series at the specified path (2-arity version).

Same as aggregate/3 but with default slices=1.

Link to this callback

aggregate(series, path, slices)

View Source 
@callback aggregate(series :: map(), path :: String.t(), slices :: integer()) :: list()

Aggregate data from a series at the specified path.

Parameters

  • series: Map containing :at (timeline) and :values (data points) keys
  • path: String path to the values to aggregate (e.g., "count", "metrics.duration")
  • slices: Number of slices for windowed aggregation (optional, defaults to 1)

Returns

List of aggregated values. For single aggregation, returns single-item list. For sliced aggregation, returns list with one item per slice.

Example 

series = %{
  at: [~U[2025-08-18 10:00:00Z], ~U[2025-08-18 11:00:00Z]],
  values: [%{"count" => 10}, %{"count" => 15}]
}

aggregate(series, "count", 1)
# => [25]  # sum of all values

aggregate(series, "count", 2) 
# => [10, 15]  # each slice separately