anomaly 0.1.0 → 0.3.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: 7cf7957fa3d0c9ddf4fb5dbc59c6b153ae0122d6777dffc4b088156a34b44c04
+  data.tar.gz: a4130a661e44bf2363ef024c8e5296803cab479d9bc69cff7661a5d702feccb7
+SHA512:
+  metadata.gz: 3c13a829fdc03e54b408b7bfc8ab0ea5ccb6b7092a12146ee092071ccf81e73da278b7910b0dd44d846b2a3619580567da5c2116968d0d73b4bd13d5df414c1c
+  data.tar.gz: e7b7aff28acbaf2c9c3a95d62ccf8c10829689896a98e32a5f85ce22114754d42eeb0e5ccf70bbf49a4fda850fadcb367747caabc41b944db5e82bfd7686f351

data/CHANGELOG.md

@@ -0,0 +1,18 @@
+## 0.3.0 (2022-09-05)
+
+- Dropped support for `narray` (use `numo-narray` instead)
+- Dropped support for Ruby < 2.7
+
+## 0.2.1 (2020-04-16)
+
+- Added support for multiple predictions
+
+## 0.2.0 (2019-10-27)
+
+- Switched to Ruby `sum` for performance
+- Added support for Numo::NArray
+- Use keyword arguments
+
+## 0.1.0 (2011-12-18)
+
+- Started changelog

data/{LICENSE → LICENSE.txt}

@@ -1,4 +1,4 @@
-Copyright (c) 2011 Andrew Kane
+Copyright (c) 2011-2022 Andrew Kane
 
 MIT License
 
@@ -19,4 +19,4 @@ MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
 LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
 OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

data/README.md

@@ -1,32 +1,20 @@
 # Anomaly
 
-Easy-to-use anomaly detection
+Easy-to-use anomaly detection for Ruby
+
+[![Build Status](https://github.com/ankane/anomaly/workflows/build/badge.svg?branch=master)](https://github.com/ankane/anomaly/actions)
 
 ## Installation
 
-Add this line to your application's Gemfile:
+Add this line to your application’s Gemfile:
 
 ```ruby
 gem "anomaly"
 ```
 
-And then execute:
+## Getting Started
 
-```sh
-bundle install
-```
-
-For max performance (trains ~3x faster for large datasets), also install the NArray gem:
-
-```ruby
-gem "narray"
-```
-
-Anomaly will automatically detect it and use it.
-
-## How to Use
-
-Say we have weather data and we want to predict if it's sunny. In this example, sunny days are non-anomalies, and days with other types of weather (rain, snow, etc.) are anomalies. The data looks like:
+Say we have weather data and we want to predict if it’s sunny. In this example, sunny days are non-anomalies, and days with other types of weather (rain, snow, etc.) are anomalies. The data looks like:
 
 ```ruby
 # [temperature(°F), humidity(%), pressure(in), sunny?(y=0, n=1)]
@@ -41,57 +29,78 @@ weather_data = [
 
 The last column **must** be 0 for non-anomalies, 1 for anomalies. Non-anomalies are used to train the detector, and both anomalies and non-anomalies are used to find the best value of ε.
 
-To train the detector and test for anomalies, run:
+Train the detector
 
 ```ruby
-ad = Anomaly::Detector.new(weather_data)
+detector = Anomaly::Detector.new(weather_data)
+```
+
+Test for anomalies
 
+```ruby
 # 85°F, 42% humidity, 12.3 in. pressure
-ad.anomaly?([85, 42, 12.3])
-# => true
+detector.anomaly?([85, 42, 12.3])
 ```
 
 Anomaly automatically finds the best value for ε, which you can access with:
 
 ```ruby
-ad.eps
+detector.eps
 ```
 
 If you already know you want ε = 0.01, initialize the detector with:
 
 ```ruby
-ad = Anomaly::Detector.new(weather_data, {:eps => 0.01})
+detector = Anomaly::Detector.new(weather_data, eps: 0.01)
 ```
 
-### Persistence
+## Persistence
 
-You can easily persist the detector to a file or database - it's very tiny.
+You can easily persist the detector to a file or database - it’s very tiny.
 
 ```ruby
-serialized_ad = Marshal.dump(ad)
-
-# Save to a file
-File.open("anomaly_detector.dump", "w") {|f| f.write(serialized_ad) }
+bin = Marshal.dump(detector)
+File.binwrite("detector.bin", bin)
+```
 
-# ...
+Then read it later:
 
-# Read it later
-ad2 = Marshal.load(File.open("anomaly_detector.dump", "r").read)
+```ruby
+bin = File.binread("detector.bin")
+detector = Marshal.load(bin)
 ```
 
-## TODO
+## Related Projects
+
+- [AnomalyDetection.rb](https://github.com/ankane/AnomalyDetection.rb) - Time series anomaly detection for Ruby
+- [Prophet.rb](https://github.com/ankane/prophet-ruby) - Time series forecasting (and anomaly detection) for Ruby
+- [IsoTree](https://github.com/ankane/isotree-ruby) - Outlier/anomaly detection for Ruby using Isolation Forest
+- [OutlierTree](https://github.com/ankane/outliertree-ruby) - Explainable outlier/anomaly detection for Ruby
+- [MIDAS](https://github.com/ankane/midas-ruby) - Edge stream anomaly detection for Ruby
+- [Trend](https://github.com/ankane/trend-ruby) - Anomaly detection and forecasting for Ruby
+
+## Credits
 
-- Train in chunks (for very large datasets)
-- Multivariate normal distribution (possibly)
+A special thanks to [Andrew Ng](https://www.coursera.org/learn/machine-learning).
+
+## History
+
+View the [changelog](https://github.com/ankane/anomaly/blob/master/CHANGELOG.md)
 
 ## Contributing
 
-1. Fork it
-2. Create your feature branch (`git checkout -b my-new-feature`)
-3. Commit your changes (`git commit -am 'Added some feature'`)
-4. Push to the branch (`git push origin my-new-feature`)
-5. Create new Pull Request
+Everyone is encouraged to help improve this project. Here are a few ways you can help:
+
+- [Report bugs](https://github.com/ankane/anomaly/issues)
+- Fix bugs and [submit pull requests](https://github.com/ankane/anomaly/pulls)
+- Write, clarify, or fix documentation
+- Suggest or add new features
 
-## Thanks
+To get started with development:
 
-A special thanks to [Andrew Ng](http://www.ml-class.org).
+```sh
+git clone https://github.com/ankane/anomaly.git
+cd anomaly
+bundle install
+bundle exec rake spec
+```

data/lib/anomaly/detector.rb

@@ -1,13 +1,18 @@
 module Anomaly
   class Detector
+    attr_reader :mean, :std
     attr_accessor :eps
 
-    def initialize(examples = nil, opts = {})
+    def initialize(examples = nil, **opts)
       @m = 0
-      train(examples, opts) if examples
+      train(examples, **opts) if examples
     end
 
-    def train(examples, opts = {})
+    def train(examples, eps: 0)
+      # for Numo::NArray
+      # TODO make more efficient when possible
+      examples = examples.to_a
+
       raise "No examples" if examples.empty?
       raise "Must have at least two columns" if examples.first.size < 2
 
@@ -24,7 +29,7 @@ module Anomaly
 
       raise "Must have at least one non-anomaly" if non_anomalies.empty?
 
-      @eps = (opts[:eps] || 0).to_f
+      @eps = eps
       if @eps > 0
         # Use all non-anomalies to train.
         training_examples = non_anomalies
@@ -33,28 +38,33 @@ module Anomaly
         test_examples.concat(anomalies)
       end
       # Remove last column.
-      training_examples = training_examples.map{|e| e[0..-2]}
+      training_examples = training_examples.map { |e| e[0..-2] }
       @m = training_examples.size
       @n = training_examples.first.size
 
-      if defined?(NMatrix)
+      if defined?(Numo::SFloat)
+        training_examples = Numo::SFloat.cast(training_examples)
+        # Convert these to an Array for Marshal.dump
+        @mean = training_examples.mean(0).to_a
+        @std = training_examples.stddev(0).to_a
+      elsif defined?(NMatrix)
         training_examples = NMatrix.to_na(training_examples)
         # Convert these to an Array for Marshal.dump
         @mean = training_examples.mean(1).to_a
         @std = training_examples.stddev(1).to_a
       else
         # Default to Array, since built-in Matrix does not give us a big performance advantage.
-        cols = @n.times.map{|i| training_examples.map{|r| r[i]}}
-        @mean = cols.map{|c| mean(c)}
-        @std = cols.each_with_index.map{|c,i| std(c, @mean[i])}
+        cols = @n.times.map { |i| training_examples.map { |r| r[i] } }
+        @mean = cols.map { |c| alt_mean(c) }
+        @std = cols.each_with_index.map { |c, i| alt_std(c, @mean[i]) }
       end
-      @std.map!{|std| (std == 0 or std.nan?) ? Float::MIN : std}
+      @std.map! { |std| (std == 0 || std.nan?) ? 1e-10 : std }
 
       if @eps == 0
         # Find the best eps.
-        epss = (1..9).map{|i| [1,3,5,7,9].map{|j| (j*10**(-i)).to_f }}.flatten
-        f1_scores = epss.map{|eps| [eps, compute_f1_score(test_examples, eps)] }
-        @eps, best_f1 = f1_scores.max_by{|v| v[1]}
+        epss = (1..9).map { |i| [1, 3, 5, 7, 9].map { |j| (j * 10**(-i)).to_f } }.flatten
+        f1_scores = epss.map { |eps| [eps, compute_f1_score(test_examples, eps)] }
+        @eps, _ = f1_scores.max_by { |v| v[1] }
       end
     end
 
@@ -64,25 +74,44 @@ module Anomaly
 
     # Limit the probability of features to [0,1]
     # to keep probabilities at same scale.
+    # Use log to prevent underflow
     def probability(x)
       raise "Train me first" unless trained?
-      raise ArgumentError, "First argument must have #{@n} elements" if x.size != @n
-      @n.times.map do |i|
-        p = normal_pdf(x[i], @mean[i], @std[i])
-        (p.nan? or p > 1) ? 1 : p
-      end.reduce(1, :*)
+
+      singular = !x.first.is_a?(Array)
+      x = [x] if singular
+
+      y =
+        x.map do |xi|
+          prob = 0
+          @n.times.map do |i|
+            pi = normal_pdf(xi[i], @mean[i], @std[i])
+            prob += Math.log(pi > 1 ? 1 : pi)
+          end
+          Math.exp(prob)
+        end
+
+      singular ? y.first : y
     end
 
     def anomaly?(x, eps = @eps)
-      probability(x) < eps
+      y = probability(x)
+
+      if y.is_a?(Array)
+        y.map do |yi|
+          yi < eps
+        end
+      else
+        y < eps
+      end
     end
 
     protected
 
-    SQRT2PI = Math.sqrt(2*Math::PI)
+    SQRT2PI = Math.sqrt(2 * Math::PI)
 
     def normal_pdf(x, mean = 0, std = 1)
-      1/(SQRT2PI*std)*Math.exp(-((x - mean)**2/(2.0*(std**2))))
+      1 / (SQRT2PI * std) * Math.exp(-((x - mean)**2 / (2.0 * (std**2))))
     end
 
     # Find best eps.
@@ -99,8 +128,8 @@ module Anomaly
       fn = 0
       examples.each do |example|
         act = example.last != 0
-        pred = self.anomaly?(example[0..-2], eps)
-        if act and pred
+        pred = anomaly?(example[0..-2], eps)
+        if act && pred
           tp += 1
         elsif pred # and !act
           fp += 1
@@ -120,13 +149,12 @@ module Anomaly
 
     # Not used for NArray
 
-    def mean(x)
-      x.inject(0.0){|a, i| a + i}/x.size
+    def alt_mean(x)
+      x.sum / x.size
     end
 
-    def std(x, mean)
-      Math.sqrt(x.inject(0.0){|a, i| a + (i - mean) ** 2}/(x.size - 1))
+    def alt_std(x, mean)
+      Math.sqrt(x.sum { |i| (i - mean)**2 }.to_f / (x.size - 1))
     end
-
   end
 end

data/lib/anomaly/version.rb

@@ -1,3 +1,3 @@
 module Anomaly
-  VERSION = "0.1.0"
+  VERSION = "0.3.0"
 end

metadata

@@ -1,98 +1,48 @@
 --- !ruby/object:Gem::Specification
 name: anomaly
 version: !ruby/object:Gem::Version
-  version: 0.1.0
-  prerelease: 
+  version: 0.3.0
 platform: ruby
 authors:
 - Andrew Kane
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2011-12-19 00:00:00.000000000Z
-dependencies:
-- !ruby/object:Gem::Dependency
-  name: rake
-  requirement: &2155813680 !ruby/object:Gem::Requirement
-    none: false
-    requirements:
-    - - ! '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-  type: :development
-  prerelease: false
-  version_requirements: *2155813680
-- !ruby/object:Gem::Dependency
-  name: rspec
-  requirement: &2155813180 !ruby/object:Gem::Requirement
-    none: false
-    requirements:
-    - - ! '>='
-      - !ruby/object:Gem::Version
-        version: 2.0.0
-  type: :development
-  prerelease: false
-  version_requirements: *2155813180
-- !ruby/object:Gem::Dependency
-  name: narray
-  requirement: &2155812760 !ruby/object:Gem::Requirement
-    none: false
-    requirements:
-    - - ! '>='
-      - !ruby/object:Gem::Version
-        version: '0'
-  type: :development
-  prerelease: false
-  version_requirements: *2155812760
-description: Easy-to-use anomaly detection
-email:
-- andrew@getformidable.com
+date: 2022-09-05 00:00:00.000000000 Z
+dependencies: []
+description:
+email: andrew@ankane.org
 executables: []
 extensions: []
 extra_rdoc_files: []
 files:
-- .gitignore
-- .rspec
-- Gemfile
-- LICENSE
+- CHANGELOG.md
+- LICENSE.txt
 - README.md
-- Rakefile
-- anomaly.gemspec
 - lib/anomaly.rb
 - lib/anomaly/detector.rb
 - lib/anomaly/version.rb
-- spec/anomaly/detector_spec.rb
-- spec/spec_helper.rb
 homepage: https://github.com/ankane/anomaly
-licenses: []
-post_install_message: 
+licenses:
+- MIT
+metadata: {}
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
-  none: false
   requirements:
-  - - ! '>='
+  - - ">="
     - !ruby/object:Gem::Version
-      version: '0'
-      segments:
-      - 0
-      hash: 1886385059125072633
+      version: '2.7'
 required_rubygems_version: !ruby/object:Gem::Requirement
-  none: false
   requirements:
-  - - ! '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
-      segments:
-      - 0
-      hash: 1886385059125072633
 requirements: []
-rubyforge_project: 
-rubygems_version: 1.8.11
-signing_key: 
-specification_version: 3
-summary: Easy-to-use anomaly detection
-test_files:
-- spec/anomaly/detector_spec.rb
-- spec/spec_helper.rb
+rubygems_version: 3.3.7
+signing_key:
+specification_version: 4
+summary: Easy-to-use anomaly detection for Ruby
+test_files: []

data/.gitignore

@@ -1,17 +0,0 @@
-*.gem
-*.rbc
-.bundle
-.config
-.yardoc
-Gemfile.lock
-InstalledFiles
-_yardoc
-coverage
-doc/
-lib/bundler/man
-pkg
-rdoc
-spec/reports
-test/tmp
-test/version_tmp
-tmp

data/.rspec

@@ -1 +0,0 @@
---color

data/Gemfile

@@ -1,4 +0,0 @@
-source 'https://rubygems.org'
-
-# Specify your gem's dependencies in anomaly.gemspec
-gemspec

data/Rakefile

@@ -1,25 +0,0 @@
-#!/usr/bin/env rake
-require "bundler/gem_tasks"
-require "rspec/core/rake_task"
-RSpec::Core::RakeTask.new("spec")
-
-require "benchmark"
-require "anomaly"
-
-task :benchmark do
-  examples = 1_000_000.times.map{ [rand, rand, rand, 0] }
-
-  Benchmark.bm do |x|
-    x.report { Anomaly::Detector.new(examples, {:eps => 0.5}) }
-    require "narray"
-    x.report { Anomaly::Detector.new(examples, {:eps => 0.5}) }
-  end
-end
-
-task :random_examples do
-  examples = 10_000.times.map{ [rand, rand(10), rand(100), 0] } +
-    100.times.map{ [rand + 1, rand(10) + 2, rand(100) + 20, 1] }
-
-  ad = Anomaly::Detector.new(examples)
-  puts ad.eps
-end

data/anomaly.gemspec

@@ -1,21 +0,0 @@
-# -*- encoding: utf-8 -*-
-require File.expand_path('../lib/anomaly/version', __FILE__)
-
-Gem::Specification.new do |gem|
-  gem.authors       = ["Andrew Kane"]
-  gem.email         = ["andrew@getformidable.com"]
-  gem.description   = %q{Easy-to-use anomaly detection}
-  gem.summary       = %q{Easy-to-use anomaly detection}
-  gem.homepage      = "https://github.com/ankane/anomaly"
-
-  gem.executables   = `git ls-files -- bin/*`.split("\n").map{ |f| File.basename(f) }
-  gem.files         = `git ls-files`.split("\n")
-  gem.test_files    = `git ls-files -- {test,spec,features}/*`.split("\n")
-  gem.name          = "anomaly"
-  gem.require_paths = ["lib"]
-  gem.version       = Anomaly::VERSION
-
-  gem.add_development_dependency "rake"
-  gem.add_development_dependency "rspec", ">= 2.0.0"
-  gem.add_development_dependency "narray"
-end

data/spec/anomaly/detector_spec.rb

@@ -1,71 +0,0 @@
-require "spec_helper"
-
-describe Anomaly::Detector do
-  let(:examples) { [[-1,-2,0],[0,0,0],[1,2,0]] }
-  let(:ad) { Anomaly::Detector.new(examples) }
-
-  # mean = [0, 0], std = [1, 2]
-  it "computes the right probability" do
-    ad.probability([0,0]).should == 0.079577471545947667
-  end
-
-  it "marshalizes" do
-    expect{ Marshal.dump(ad) }.to_not raise_error
-  end
-
-  context "when standard deviation is 0" do
-    let(:examples) { [[0,0],[0,0]] }
-
-    it "returns infinity for mean" do
-      ad.probability([0]).should == 1
-    end
-
-    it "returns 0 for not mean" do
-      ad.probability([1]).should == 0
-    end
-  end
-
-  context "when examples is an array" do
-    let(:examples) { [[-1,-2,0],[0,0,0],[1,2,0]] }
-    let(:sample) { [rand, rand] }
-
-    it "returns the same probability as an NMatrix" do
-      prob = ad.probability(sample)
-      Object.send(:remove_const, :NMatrix)
-      prob.should == Anomaly::Detector.new(examples).probability(sample)
-    end
-  end
-
-  context "when lots of samples" do
-    let(:examples) { m.times.map{[0,0]} }
-    let(:m) { rand(100) + 1 }
-
-    it { ad.trained?.should be_true }
-  end
-
-  context "when no samples" do
-    let(:examples) { nil }
-
-    it { ad.trained?.should be_false }
-  end
-
-  context "when pdf is greater than 1" do
-    let(:examples) { 100.times.map{[0,0]}.push([1,0]) }
-
-    it { ad.probability([0]).should == 1 }
-  end
-
-  context "when only anomalies" do
-    let(:examples) { [[0,1]] }
-
-    it "raises error" do
-      expect{ ad }.to raise_error RuntimeError, "Must have at least one non-anomaly"
-    end
-  end
-
-  context "when only one non-anomaly" do
-    let(:examples) { [[0,0]] }
-
-    it { ad.eps.should == 1e-1 }
-  end
-end

data/spec/spec_helper.rb

@@ -1,8 +0,0 @@
-require "rubygems"
-require "bundler/setup"
-
-require "anomaly"
-require "narray"
-
-RSpec.configure do |config|
-end