vowpalwabbit 0.1.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: fd6a68e5d6103be4f6f58bd3fb6cb9785bd47382d7787b8d0ebc2c0336d87af3
+  data.tar.gz: 2f290dfdfe0539b4daf3e73dfbb39d04fa97e2f5c9aec08d43b0ee81db1d49b2
+SHA512:
+  metadata.gz: 7e63904cf470666bbec00e9096e1701c1824af4b7790aa9d137a8a779eefb6d6833aadf4043f7d941e2f8395dfda532349bada33ffaa5cfa8461a593b9e80088
+  data.tar.gz: 0d6b47d922da2cfefb525f59e718e6f3be4c7e99033647f0f8814b9efde0095399b1b87c5922143cefeb8f0b7207ca4811b8c1c29505ea7b5af8cdc5c9910acc

data/CHANGELOG.md

@@ -0,0 +1,3 @@
+## 0.1.0
+
+- First release

data/LICENSE.txt

@@ -0,0 +1,22 @@
+Copyright (c) 2019 Andrew Kane
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+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.

data/README.md

@@ -0,0 +1,125 @@
+# Vowpal Wabbit
+
+[Vowpal Wabbit](https://vowpalwabbit.org) - fast online machine learning - for Ruby
+
+:fire: Uses the C API for blazing performance
+
+## Installation
+
+First, [install Vowpal Wabbit](https://vowpalwabbit.org/start.html). For Homebrew, use:
+
+```sh
+brew install vowpal-wabbit
+```
+
+Add this line to your application’s Gemfile:
+
+```ruby
+gem 'vowpalwabbit'
+```
+
+## Getting Started
+
+Prep your data
+
+```ruby
+x = [[1, 2], [3, 4], [5, 6], [7, 8]]
+y = [1, 2, 3, 4]
+```
+
+Train a model
+
+```ruby
+model = VowpalWabbit::Regressor.new(l: 100)
+model.fit(x, y)
+```
+
+Use `VowpalWabbit::Classifier` for classification and `VowpalWabbit::Model` for other models
+
+Make predictions
+
+```ruby
+model.predict(x)
+```
+
+Save the model to a file
+
+```ruby
+model.save_model("model.bin")
+```
+
+Load the model from a file
+
+```ruby
+model.load_model("model.bin")
+```
+
+Train online
+
+```ruby
+model.partial_fit(x, y)
+```
+
+Get the intercept and coefficients
+
+```ruby
+model.intercept
+model.coefs
+```
+
+Score - R-squared for regression and accuracy for classification
+
+```ruby
+model.score(x, y)
+```
+
+## Parameters
+
+Specify parameters
+
+```ruby
+model = VowpalWabbit::Model.new(cb: 4)
+```
+
+Supports the same parameters as the [CLI](https://github.com/VowpalWabbit/vowpal_wabbit/wiki/Command-Line-Arguments)
+
+## Data
+
+Data can be an array of arrays
+
+```ruby
+[[1, 2, 3], [4, 5, 6]]
+```
+
+Or an array of strings
+
+```ruby
+[
+  "0 | price:.23 sqft:.25 age:.05 2006",
+  "1 2 'second_house | price:.18 sqft:.15 age:.35 1976",
+  "0 1 0.5 'third_house | price:.53 sqft:.32 age:.87 1924"
+]
+```
+
+Or a path to a file
+
+```ruby
+model.fit("train.txt")
+model.predict("train.txt")
+model.score("train.txt")
+```
+
+Read more about the [input format](https://github.com/VowpalWabbit/vowpal_wabbit/wiki/Input-format)
+
+## History
+
+View the [changelog](https://github.com/ankane/vowpalwabbit/blob/master/CHANGELOG.md)
+
+## Contributing
+
+Everyone is encouraged to help improve this project. Here are a few ways you can help:
+
+- [Report bugs](https://github.com/ankane/vowpalwabbit/issues)
+- Fix bugs and [submit pull requests](https://github.com/ankane/vowpalwabbit/pulls)
+- Write, clarify, or fix documentation
+- Suggest or add new features

data/lib/vowpalwabbit.rb

@@ -0,0 +1,20 @@
+# dependencies
+require "ffi"
+
+# modules
+require "vowpalwabbit/model"
+require "vowpalwabbit/classifier"
+require "vowpalwabbit/regressor"
+require "vowpalwabbit/version"
+
+module VowpalWabbit
+  class Error < StandardError; end
+
+  class << self
+    attr_accessor :ffi_lib
+  end
+  self.ffi_lib = ["vw_c_wrapper"]
+
+  # friendlier error message
+  autoload :FFI, "vowpalwabbit/ffi"
+end

data/lib/vowpalwabbit/classifier.rb

@@ -0,0 +1,18 @@
+module VowpalWabbit
+  class Classifier < Model
+    def initialize(**params)
+      super({loss_function: "logistic"}.merge(params))
+    end
+
+    def predict(x)
+      predictions = super
+      predictions.map { |v| v >= 0 ? 1 : -1 }
+    end
+
+    def score(x, y = nil)
+      y_pred, y = predict_for_score(x, y)
+      y_pred.map! { |v| v >= 0 ? 1 : -1 }
+      y_pred.zip(y).select { |yp, yt| yp == yt }.count / y.count.to_f
+    end
+  end
+end

data/lib/vowpalwabbit/ffi.rb

@@ -0,0 +1,41 @@
+module VowpalWabbit
+  module FFI
+    extend ::FFI::Library
+
+    begin
+      ffi_lib VowpalWabbit.ffi_lib
+    rescue LoadError => e
+      raise e if ENV["VOWPALWABBIT_DEBUG"]
+      raise LoadError, "Could not find Vowpal Wabbit"
+    end
+
+    # https://github.com/VowpalWabbit/vowpal_wabbit/blob/master/vowpalwabbit/vwdll.h
+    # keep same order
+
+    attach_function :VW_InitializeA, %i[string], :pointer
+    attach_function :VW_InitializeWithModel, %i[string pointer size_t], :pointer
+    attach_function :VW_SeedWithModel, %i[pointer string], :pointer
+    attach_function :VW_Finish_Passes, %i[pointer], :void
+    attach_function :VW_Finish, %i[pointer], :void
+    attach_function :VW_ImportExample, %i[pointer string pointer size_t], :pointer
+    attach_function :VW_ReadExampleA, %i[pointer string], :pointer
+    attach_function :VW_StartParser, %i[pointer], :void
+    attach_function :VW_EndParser, %i[pointer], :void
+    attach_function :VW_GetExample, %i[pointer], :pointer
+    attach_function :VW_FinishExample, %i[pointer pointer], :void
+    attach_function :VW_GetLabel, %i[pointer], :float
+    attach_function :VW_GetFeatureNumber, %i[pointer], :size_t
+    attach_function :VW_GetFeatures, %i[pointer pointer pointer], :pointer
+    attach_function :VW_HashSpaceA, %i[pointer string], :size_t
+    attach_function :VW_Learn, %i[pointer pointer], :float
+    attach_function :VW_Predict, %i[pointer pointer], :float
+    attach_function :VW_PredictCostSensitive, %i[pointer pointer], :float
+    attach_function :VW_Get_Weight, %i[pointer size_t size_t], :float
+    attach_function :VW_Set_Weight, %i[pointer size_t size_t float], :void
+    attach_function :VW_Num_Weights, %i[pointer], :size_t
+    attach_function :VW_Get_Stride, %i[pointer], :size_t
+    attach_function :VW_SaveModel, %i[pointer], :void
+    attach_function :VW_CopyModelData, %i[pointer pointer pointer pointer], :void
+    attach_function :VW_FreeIOBuf, %i[pointer], :void
+  end
+end

data/lib/vowpalwabbit/model.rb

@@ -0,0 +1,140 @@
+module VowpalWabbit
+  class Model
+    def initialize(**params)
+      # add strict parse once exceptions are handled properly
+      # https://github.com/VowpalWabbit/vowpal_wabbit/issues/2004
+      @params = {quiet: true}.merge(params)
+    end
+
+    def fit(x, y = nil)
+      @handle = nil
+      partial_fit(x, y)
+    end
+
+    def partial_fit(x, y = nil)
+      each_example(x, y) do |example|
+        FFI.VW_Learn(handle, example)
+      end
+      nil
+    end
+
+    def predict(x)
+      out = []
+      each_example(x) do |example|
+        out << predict_example(example)
+      end
+      out
+    end
+
+    def coefs
+      num_weights = FFI.VW_Num_Weights(handle)
+      coefs = {}
+      num_weights.times.map do |i|
+        weight = FFI.VW_Get_Weight(handle, i, 0)
+        coefs[i] = weight if weight != 0
+      end
+      coefs
+    end
+
+    def save_model(filename)
+      buffer_handle = ::FFI::MemoryPointer.new(:pointer)
+      output_data = ::FFI::MemoryPointer.new(:pointer)
+      output_size = ::FFI::MemoryPointer.new(:size_t)
+      FFI.VW_CopyModelData(handle, buffer_handle, output_data, output_size)
+      bin_str = output_data.read_pointer.read_string(output_size.read(:size_t))
+      FFI.VW_FreeIOBuf(buffer_handle.read_pointer)
+      File.binwrite(filename, bin_str)
+      nil
+    end
+
+    def load_model(filename)
+      bin_str = File.binread(filename)
+      model_data = ::FFI::MemoryPointer.new(:char, bin_str.bytesize)
+      model_data.put_bytes(0, bin_str)
+      @handle = FFI.VW_InitializeWithModel(param_str, model_data, bin_str.bytesize)
+      nil
+    end
+
+    private
+
+    # TODO clean-up handle
+    def handle
+      @handle ||= FFI.VW_InitializeA(param_str)
+    end
+
+    def param_str
+      args =
+        @params.map do |k, v|
+          check_param(k.to_s)
+          check_param(v.to_s)
+
+          if v == true
+            "--#{k}"
+          elsif !v
+            nil
+          elsif k.size == 1
+            "-#{k} #{v}"
+          else
+            "--#{k} #{v}"
+          end
+        end
+      args.compact.join(" ")
+    end
+
+    def check_param(v)
+      raise ArgumentError, "Invalid parameter" if /[[:space:]]/.match(v)
+    end
+
+    def predict_example(example)
+      if @params[:cb]
+        FFI.VW_PredictCostSensitive(handle, example)
+      else
+        FFI.VW_Predict(handle, example)
+      end
+    end
+
+    # get both in one pass for efficiency
+    def predict_for_score(x, y)
+      if x.is_a?(String) && !y
+        y_pred = []
+        y = []
+        each_example(x) do |example|
+          y_pred << predict_example(example)
+          y << FFI.VW_GetLabel(example)
+        end
+        [y_pred, y]
+      else
+        [predict(x), y]
+      end
+    end
+
+    # TODO support compressed files
+    def each_example(x, y = nil)
+      each_line(x, y) do |line|
+        example = FFI.VW_ReadExampleA(handle, line)
+        yield example
+        FFI.VW_FinishExample(handle, example)
+      end
+    end
+
+    def each_line(x, y)
+      if x.is_a?(String)
+        raise ArgumentError, "Cannot pass y with file" if y
+
+        File.foreach(x) do |line|
+          yield line
+        end
+      else
+        raise ArgumentError, "x and y must have same size" if y && x.size != y.size
+
+        x.zip(y || []) do |xi, yi|
+          if xi.is_a?(String)
+            yield xi
+          else
+            yield "#{yi} 1 | #{xi.map.with_index { |v, i| "#{i}:#{v}" }.join(" ")}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/vowpalwabbit/regressor.rb

@@ -0,0 +1,16 @@
+module VowpalWabbit
+  class Regressor < Model
+    def score(x, y = nil)
+      y_pred, y = predict_for_score(x, y)
+
+      # r2
+      sse = y_pred.zip(y).map { |yp, yt| (yp - yt) ** 2 }.sum
+      sst = y.map { |yi| yi ** 2 }.sum - (y.sum ** 2) / y.size
+      1 - sse / sst
+    end
+
+    def intercept
+      FFI.VW_Get_Weight(handle, 116060, 0)
+    end
+  end
+end

data/lib/vowpalwabbit/version.rb

@@ -0,0 +1,3 @@
+module VowpalWabbit
+  VERSION = "0.1.0"
+end

metadata

@@ -0,0 +1,107 @@
+--- !ruby/object:Gem::Specification
+name: vowpalwabbit
+version: !ruby/object:Gem::Version
+  version: 0.1.0
+platform: ruby
+authors:
+- Andrew Kane
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2019-10-21 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: ffi
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: bundler
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rake
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: minitest
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '5'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '5'
+description: 
+email: andrew@chartkick.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- CHANGELOG.md
+- LICENSE.txt
+- README.md
+- lib/vowpalwabbit.rb
+- lib/vowpalwabbit/classifier.rb
+- lib/vowpalwabbit/ffi.rb
+- lib/vowpalwabbit/model.rb
+- lib/vowpalwabbit/regressor.rb
+- lib/vowpalwabbit/version.rb
+homepage: https://github.com/ankane/vowpalwabbit
+licenses:
+- MIT
+metadata: {}
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '2.4'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  requirements:
+  - - ">="
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubygems_version: 3.0.3
+signing_key: 
+specification_version: 4
+summary: Vowpal Wabbit - fast online machine learning - for Ruby
+test_files: []