statistics

package

v0.1.1 Latest Latest Go to latest Published: Jul 14, 2026 License: Apache-2.0 Imports: 3 Imported by: 0

Details

Valid go.mod file
Redistributable license
Tagged version
Stable version
Learn more about best practices

Repository

github.com/whyiug/agentapi-doctor

Links

Open Source Insights

Documentation ¶

Overview ¶

Package statistics contains deterministic statistical helpers used by behavioral and operational-reliability oracles.

Index ¶

func Quantile(values []float64, probability float64) (float64, error)
type Flakiness
- func MeasureFlakiness(verdicts []string) (Flakiness, error)
type Interval
- func Wilson(successes, samples int64, z float64) (Interval, error)

Constants ¶

This section is empty.

Variables ¶

This section is empty.

Functions ¶

func Quantile ¶

func Quantile(values []float64, probability float64) (float64, error)

Quantile computes the R-7 sample quantile used by common statistical packages. The input is copied so callers never observe a reorder.

Types ¶

type Flakiness ¶

type Flakiness struct {
	Samples       int64   `json:"samples"`
	Passes        int64   `json:"passes"`
	Failures      int64   `json:"failures"`
	Inconclusive  int64   `json:"inconclusive"`
	Disagreement  float64 `json:"disagreement"`
	FlakyDetected bool    `json:"flaky_detected"`
}

Flakiness describes whether repeated completed attempts disagree. It does not replace the verdict of any individual attempt.

func MeasureFlakiness ¶

func MeasureFlakiness(verdicts []string) (Flakiness, error)

MeasureFlakiness counts stable public verdict labels. WARN is neither a pass nor target failure and therefore contributes to inconclusive samples.

type Interval ¶

type Interval struct {
	SampleCount int64   `json:"sample_count"`
	Estimate    float64 `json:"estimate"`
	LowerBound  float64 `json:"lower_bound"`
	UpperBound  float64 `json:"upper_bound"`
	Method      string  `json:"method"`
}

Interval is a bounded estimate over [0,1].

func Wilson ¶

func Wilson(successes, samples int64, z float64) (Interval, error)

Wilson returns a two-sided Wilson score interval. z=1.959963984540054 is the conventional 95% interval. Zero samples are rejected rather than being represented as an apparently meaningful 0% estimate.

Source Files ¶

View all Source files

statistics.go

?	: This menu
/	: Search site
f or F	: Jump to
y or Y	: Canonical URL