Liking cljdoc? Tell your friends :D

criterium.analyse.metrics-samples


autocorrelation-for-metricclj

(autocorrelation-for-metric metric->values outliers metric-config _options)

Compute autocorrelation analysis for a single metric's samples.

When outliers is provided, filters outlier samples before computing ACF. When outliers is nil, uses all samples (for pattern detection before outlier removal).

Returns nil for metrics with insufficient samples (n < 20).

Compute autocorrelation analysis for a single metric's samples.

When outliers is provided, filters outlier samples before computing ACF.
When outliers is nil, uses all samples (for pattern detection before
outlier removal).

Returns nil for metrics with insufficient samples (n < 20).
sourceraw docstring

classifierclj

(classifier [low-severe low-mild high-mild high-severe])
source

distribution-fit-for-metricclj

(distribution-fit-for-metric metric->values outliers metric-config options)

Compute distribution fitting for a single metric's samples.

Returns fit results including :sample-range [min max] of the filtered samples used for fitting, enabling viewers to display consistent axis ranges.

Compute distribution fitting for a single metric's samples.

Returns fit results including :sample-range [min max] of the filtered samples
used for fitting, enabling viewers to display consistent axis ranges.
sourceraw docstring

histogramclj

(histogram metric->values quantiles outliers metric-config options)
source

kde-for-metricclj

(kde-for-metric metric->values outliers metric-config options)

Compute KDE for a single metric's samples, filtering outliers if present.

Compute KDE for a single metric's samples, filtering outliers if present.
sourceraw docstring

modes-for-metricclj

(modes-for-metric kde-data samples-arr outliers metric-config options)

Compute modes with statistical validation for a single metric.

Takes KDE output and raw samples, runs multimodality test for k=1 up to max-modes, and computes confidence intervals for detected modes.

Uses early stopping: stops testing k values once p-value >= alpha (fail to reject H0: at most k modes). Higher k values would also fail to reject, so testing them is unnecessary.

Options:

  • :method - test method, :acr (default) or :silverman
  • :mode-method - mode finding method:
    • :isj (default) - find modes from KDE density at ISJ bandwidth
    • :critical - find modes at critical bandwidth for validated k modes
  • :max-modes, :n-bootstrap, :alpha, :n-points - as usual
Compute modes with statistical validation for a single metric.

Takes KDE output and raw samples, runs multimodality test for k=1 up
to max-modes, and computes confidence intervals for detected modes.

Uses early stopping: stops testing k values once p-value >=
alpha (fail to reject H0: at most k modes). Higher k values would also
fail to reject, so testing them is unnecessary.

Options:
- :method - test method, :acr (default) or :silverman
- :mode-method - mode finding method:
  - :isj (default) - find modes from KDE density at ISJ bandwidth
  - :critical - find modes at critical bandwidth for validated k modes
- :max-modes, :n-bootstrap, :alpha, :n-points - as usual
sourceraw docstring

outlier-countclj

(outlier-count low-severe low-mild high-mild high-severe)
source

samples-outliersclj

(samples-outliers metric-configs all-quantiles samples options)

Compute outliers for each metric. Returns a map with thresholds, outliers, outlier-counts, outlier-method, and (for medcouple method) medcouple for each metric path.

Options: :outlier-method - :medcouple (default) or :tukey :medcouple uses the adjusted boxplot method accounting for skewness via medcouple statistic :tukey uses Tukey's symmetric 1.5×IQR whiskers

Compute outliers for each metric.
Returns a map with thresholds, outliers, outlier-counts, outlier-method,
and (for medcouple method) medcouple for each metric path.

Options:
  :outlier-method - :medcouple (default) or :tukey
                    :medcouple uses the adjusted boxplot method accounting for
                    skewness via medcouple statistic
                    :tukey uses Tukey's symmetric 1.5×IQR whiskers
sourceraw docstring

tail-analysis-for-metricclj

(tail-analysis-for-metric metric->values metric-config options)

Compute tail analysis for a single metric's samples.

Uses raw samples WITHOUT outlier filtering - tail analysis requires the extreme values that would normally be considered outliers.

Options: :threshold - explicit threshold value for POT :threshold-quantile - quantile to use as threshold (default 0.9) :k-range - range of k values for Hill estimator :high-quantiles - quantiles to estimate (default [0.99 0.999 0.9999])

Compute tail analysis for a single metric's samples.

Uses raw samples WITHOUT outlier filtering - tail analysis requires the
extreme values that would normally be considered outliers.

Options:
  :threshold - explicit threshold value for POT
  :threshold-quantile - quantile to use as threshold (default 0.9)
  :k-range - range of k values for Hill estimator
  :high-quantiles - quantiles to estimate (default [0.99 0.999 0.9999])
sourceraw docstring

cljdoc builds & hosts documentation for Clojure/Script libraries

Keyboard shortcuts
Ctrl+kJump to recent docs
Move to previous article
Move to next article
Ctrl+/Jump to the search field
× close