From 49e6369fe732c0f81e5b03b36e345afbf3c79a15 Mon Sep 17 00:00:00 2001 From: Dmitry Vyukov Date: Wed, 24 Jul 2024 12:08:49 +0200 Subject: pkg/stat: rename package name to singular form Go package names should generally be singular form: https://go.dev/blog/package-names https://rakyll.org/style-packages https://groups.google.com/g/golang-nuts/c/buBwLar1gNw --- pkg/stats/sample/pvalue.go | 20 ----------- pkg/stats/sample/sample.go | 73 ----------------------------------------- pkg/stats/sample/sample_test.go | 66 ------------------------------------- 3 files changed, 159 deletions(-) delete mode 100644 pkg/stats/sample/pvalue.go delete mode 100644 pkg/stats/sample/sample.go delete mode 100644 pkg/stats/sample/sample_test.go (limited to 'pkg/stats/sample') diff --git a/pkg/stats/sample/pvalue.go b/pkg/stats/sample/pvalue.go deleted file mode 100644 index acfff4bc4..000000000 --- a/pkg/stats/sample/pvalue.go +++ /dev/null @@ -1,20 +0,0 @@ -// Copyright 2021 syzkaller project authors. All rights reserved. -// Use of this source code is governed by Apache 2 LICENSE that can be found in the LICENSE file. - -package sample - -// TODO: I didn't find the substitution as of Feb 2023. Let's keep it as is while it works. -import "golang.org/x/perf/benchstat" // nolint:all - -// Mann-Whitney U test. -func UTest(old, new *Sample) (pval float64, err error) { - // Unfortunately we cannot just invoke MannWhitneyUTest from x/perf/benchstat/internal/stats, - // so we first wrap the data in Metrics. - mOld := benchstat.Metrics{ - RValues: old.Xs, - } - mNew := benchstat.Metrics{ - RValues: new.Xs, - } - return benchstat.UTest(&mOld, &mNew) -} diff --git a/pkg/stats/sample/sample.go b/pkg/stats/sample/sample.go deleted file mode 100644 index 740f9aefe..000000000 --- a/pkg/stats/sample/sample.go +++ /dev/null @@ -1,73 +0,0 @@ -// Copyright 2021 syzkaller project authors. All rights reserved. -// Use of this source code is governed by Apache 2 LICENSE that can be found in the LICENSE file. - -// Package sample provides various statistical operations and algorithms. -package sample - -import ( - "math" - "sort" -) - -// Sample represents a single sample - set of data points collected during an experiment. -type Sample struct { - Xs []float64 - Sorted bool -} - -func (s *Sample) Percentile(p float64) float64 { - s.Sort() - // The code below is taken from golang.org/x/perf/internal/stats - // Unfortunately, that package is internal and we cannot just import and use it. - N := float64(len(s.Xs)) - n := 1/3.0 + p*(N+1/3.0) // R8 - kf, frac := math.Modf(n) - k := int(kf) - if k <= 0 { - return s.Xs[0] - } else if k >= len(s.Xs) { - return s.Xs[len(s.Xs)-1] - } - return s.Xs[k-1] + frac*(s.Xs[k]-s.Xs[k-1]) -} - -func (s *Sample) Median() float64 { - return s.Percentile(0.5) -} - -// Remove outliers by the Tukey's fences method. -func (s *Sample) RemoveOutliers() *Sample { - if len(s.Xs) < 4 { - // If the data set is too small, we cannot reliably detect outliers anyway. - return s.Copy() - } - s.Sort() - Q1 := s.Percentile(0.25) - Q3 := s.Percentile(0.75) - minValue := Q1 - 1.5*(Q3-Q1) - maxValue := Q3 + 1.5*(Q3-Q1) - xs := []float64{} - for _, value := range s.Xs { - if value >= minValue && value <= maxValue { - xs = append(xs, value) - } - } - return &Sample{ - Xs: xs, - Sorted: s.Sorted, - } -} - -func (s *Sample) Copy() *Sample { - return &Sample{ - Xs: append([]float64{}, s.Xs...), - Sorted: s.Sorted, - } -} - -func (s *Sample) Sort() { - if !s.Sorted { - sort.Slice(s.Xs, func(i, j int) bool { return s.Xs[i] < s.Xs[j] }) - s.Sorted = true - } -} diff --git a/pkg/stats/sample/sample_test.go b/pkg/stats/sample/sample_test.go deleted file mode 100644 index ac7845ccf..000000000 --- a/pkg/stats/sample/sample_test.go +++ /dev/null @@ -1,66 +0,0 @@ -// Copyright 2021 syzkaller project authors. All rights reserved. -// Use of this source code is governed by Apache 2 LICENSE that can be found in the LICENSE file. - -package sample - -import ( - "reflect" - "testing" -) - -func TestMedian(t *testing.T) { - tests := []struct { - input []float64 - minMedian float64 - maxMedian float64 - }{ - { - input: []float64{1, 2, 3}, - minMedian: 1.99, // we cannot do exact floating point equality comparison - maxMedian: 2.01, - }, - { - input: []float64{0, 1, 2, 3}, - minMedian: 1.0, - maxMedian: 2.0, - }, - } - for _, test := range tests { - sample := Sample{Xs: test.input} - median := sample.Median() - if median < test.minMedian || median > test.maxMedian { - t.Errorf("sample %v, median got %v, median expected [%v;%v]", - test.input, median, test.minMedian, test.maxMedian) - } - } -} - -func TestRemoveOutliers(t *testing.T) { - // Some tests just to check the overall sanity of the method. - tests := []struct { - input []float64 - output []float64 - }{ - { - input: []float64{-20, 1, 2, 3, 4, 5}, - output: []float64{1, 2, 3, 4, 5}, - }, - { - input: []float64{1, 2, 3, 4, 25}, - output: []float64{1, 2, 3, 4}, - }, - { - input: []float64{-10, -5, 0, 5, 10, 15}, - output: []float64{-10, -5, 0, 5, 10, 15}, - }, - } - for _, test := range tests { - sample := Sample{Xs: test.input} - result := sample.RemoveOutliers() - result.Sort() - if !reflect.DeepEqual(result.Xs, test.output) { - t.Errorf("input: %v, expected no outliers: %v, got: %v", - test.input, test.output, result.Xs) - } - } -} -- cgit mrf-deployment