svmkit 0.1.0 → 0.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 62fdc5c03a044a7625bf2374159cf84ef32a6869
-  data.tar.gz: 4cd1c86a344cd1410a3a5c0f4bdebd04d80e7e7b
+  metadata.gz: 1d3f4f2c398faa6b9e936ec813dac3023d9d1728
+  data.tar.gz: 6fe4f9b6ac5a656fb1a3f49662997d0ebbdd8d4b
 SHA512:
-  metadata.gz: 1b704e536e183f881e6f16895ccdc1620dc8b694db7b44772db669e579ed07652df16c8de88794c65b5eeca0eeb805c415f1e44c36446cad3bdd230f3354b320
-  data.tar.gz: e25ca447621cef29ea1807168cbe6e7210308549a298db7c8797d54e127bfb1e7fe7de3c3e9a9d719cee6de1100705e352a3b76a5127282441027fd1b389e2e1
+  metadata.gz: fb1ac8798124f25cdd4dd0738dd856c8bae7e87aacac260a8ea8b1fb7388e3a966045f2382e48f241292111312eb7cb6cd69035010ee1487645f725f364ee16b
+  data.tar.gz: 99bdb17d5a2d2825e904ce2e788e31d100e1850b1d9bfc32f7e7cc48ba1b13da59b667d1ed117a6768d852d5c15d3a4c3132994bc13350315c8b07016bcbcd41

data/README.md

@@ -36,7 +36,7 @@ transformer = SVMKit::KernelApproximation::RBF.new(gamma: 2.0, n_components: 102
 transformed = transformer.fit_transform(normalized)
 
 base_classifier =
-  SVMKit::LinearModel::PegasosSVC.new(penalty: 1.0, max_iter: 50, batch_size: 20, random_seed: 1)
+  SVMKit::LinearModel::PegasosSVC.new(reg_param: 1.0, max_iter: 50, batch_size: 20, random_seed: 1)
 classifier = SVMKit::Multiclass::OneVsRestClassifier.new(estimator: base_classifier)
 classifier.fit(transformed, labels)
 

data/lib/svmkit.rb

@@ -10,6 +10,7 @@ require 'svmkit/base/classifier'
 require 'svmkit/base/transformer'
 require 'svmkit/kernel_approximation/rbf'
 require 'svmkit/linear_model/pegasos_svc'
+require 'svmkit/linear_model/logistic_regression'
 require 'svmkit/multiclass/one_vs_rest_classifier'
 require 'svmkit/preprocessing/l2_normalizer'
 require 'svmkit/preprocessing/min_max_scaler'

data/lib/svmkit/linear_model/logistic_regression.rb

@@ -0,0 +1,162 @@
+require 'svmkit/base/base_estimator'
+require 'svmkit/base/classifier'
+
+module SVMKit
+  # This module consists of the classes that implement generalized linear models.
+  module LinearModel
+    # LogisticRegression is a class that implements Logistic Regression with stochastic gradient descent (SGD) optimization.
+    # Note that the Logistic Regression of SVMKit performs as a binary classifier.
+    #
+    #   estimator =
+    #     SVMKit::LinearModel::LogisticRegression.new(reg_param: 1.0, max_iter: 100, batch_size: 20, random_seed: 1)
+    #   estimator.fit(training_samples, traininig_labels)
+    #   results = estimator.predict(testing_samples)
+    #
+    # * *Reference*:
+    #   - S. Shalev-Shwartz, Y. Singer, N. Srebro, and A. Cotter, "Pegasos: Primal Estimated sub-GrAdient SOlver for SVM," Mathematical Programming, vol. 127 (1), pp. 3--30, 2011.
+    #
+    class LogisticRegression
+      include Base::BaseEstimator
+      include Base::Classifier
+
+      DEFAULT_PARAMS = { # :nodoc:
+        reg_param: 1.0,
+        max_iter: 100,
+        batch_size: 50,
+        random_seed: nil
+      }.freeze
+
+      # The weight vector for Logistic Regression.
+      attr_reader :weight_vec
+
+      # The random generator for performing random sampling in the SGD optimization.
+      attr_reader :rng
+
+      # Create a new classifier with Logisitc Regression by the SGD optimization.
+      #
+      # :call-seq:
+      #   new(reg_param: 1.0, max_iter: 100, batch_size: 50, random_seed: 1) -> LogisiticRegression
+      #
+      # * *Arguments* :
+      #   - +:reg_param+ (Float) (defaults to: 1.0) -- The regularization parameter.
+      #   - +:max_iter+ (Integer) (defaults to: 100) -- The maximum number of iterations.
+      #   - +:batch_size+ (Integer) (defaults to: 50) -- The size of the mini batches.
+      #   - +:random_seed+ (Integer) (defaults to: nil) -- The seed value using to initialize the random generator.
+      def initialize(params = {})
+        self.params = DEFAULT_PARAMS.merge(Hash[params.map { |k, v| [k.to_sym, v] }])
+        self.params[:random_seed] ||= srand
+        @weight_vec = nil
+        @rng = Random.new(self.params[:random_seed])
+      end
+
+      # Fit the model with given training data.
+      #
+      # :call-seq:
+      #   fit(x, y) -> LogisticRegression
+      #
+      # * *Arguments* :
+      #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- The training data to be used for fitting the model.
+      #   - +y+ (NMatrix, shape: [1, n_samples]) -- The categorical variables (e.g. labels) to be used for fitting the model.
+      # * *Returns* :
+      #   - The learned classifier itself.
+      def fit(x, y)
+        # Generate binary labels
+        negative_label = y.uniq.sort.shift
+        bin_y = y.to_flat_a.map { |l| l != negative_label ? 1 : 0 }
+        # Initialize some variables.
+        n_samples, n_features = x.shape
+        rand_ids = [*0..n_samples - 1].shuffle(random: @rng)
+        @weight_vec = NMatrix.zeros([1, n_features])
+        # Start optimization.
+        params[:max_iter].times do |t|
+          # random sampling
+          subset_ids = rand_ids.shift(params[:batch_size])
+          rand_ids.concat(subset_ids)
+          # update the weight vector.
+          eta = 1.0 / (params[:reg_param] * (t + 1))
+          mean_vec = NMatrix.zeros([1, n_features])
+          subset_ids.each do |n|
+            z = @weight_vec.dot(x.row(n).transpose)[0]
+            coef = bin_y[n] / (1.0 + Math.exp(bin_y[n] * z))
+            mean_vec += x.row(n) * coef
+          end
+          mean_vec *= eta / params[:batch_size]
+          @weight_vec = @weight_vec * (1.0 - eta * params[:reg_param]) + mean_vec
+          # scale the weight vector.
+          scaler = (1.0 / params[:reg_param]**0.5) / @weight_vec.norm2
+          @weight_vec *= [1.0, scaler].min
+        end
+        self
+      end
+
+      # Calculate confidence scores for samples.
+      #
+      # :call-seq:
+      #   decision_function(x) -> NMatrix, shape: [1, n_samples]
+      #
+      # * *Arguments* :
+      #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- The samples to compute the scores.
+      # * *Returns* :
+      #   - Confidence score per sample.
+      def decision_function(x)
+        w = (@weight_vec.dot(x.transpose) * -1.0).exp + 1.0
+        w.map { |v| 1.0 / v }
+      end
+
+      # Predict class labels for samples.
+      #
+      # :call-seq:
+      #   predict(x) -> NMatrix, shape: [1, n_samples]
+      #
+      # * *Arguments* :
+      #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- The samples to predict the labels.
+      # * *Returns* :
+      #   - Predicted class label per sample.
+      def predict(x)
+        decision_function(x).map { |v| v >= 0.5 ? 1 : -1 }
+      end
+
+      # Predict probability for samples.
+      #
+      # :call-seq:
+      #   predict_proba(x) -> NMatrix, shape: [1, n_samples]
+      #
+      # * *Arguments* :
+      #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- The samples to predict the probailities.
+      # * *Returns* :
+      #   - Predicted probability per sample.
+      def predict_proba(x)
+        decision_function(x)
+      end
+
+      # Claculate the mean accuracy of the given testing data.
+      #
+      # :call-seq:
+      #   score(x, y) -> Float
+      #
+      # * *Arguments* :
+      #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- Testing data.
+      #   - +y+ (NMatrix, shape: [1, n_samples]) -- True labels for testing data.
+      # * *Returns* :
+      #   - Mean accuracy
+      def score(x, y)
+        p = predict(x)
+        n_hits = (y.to_flat_a.map.with_index { |l, n| l == p[n] ? 1 : 0 }).inject(:+)
+        n_hits / y.size.to_f
+      end
+
+      # Serializes object through Marshal#dump.
+      def marshal_dump # :nodoc:
+        { params: params, weight_vec: Utils.dump_nmatrix(@weight_vec), rng: @rng }
+      end
+
+      # Deserialize object through Marshal#load.
+      def marshal_load(obj) # :nodoc:
+        self.params = obj[:params]
+        @weight_vec = Utils.restore_nmatrix(obj[:weight_vec])
+        @rng = obj[:rng]
+        nil
+      end
+    end
+  end
+end

data/lib/svmkit/linear_model/pegasos_svc.rb

@@ -118,7 +118,7 @@ module SVMKit
       # Claculate the mean accuracy of the given testing data.
       #
       # :call-seq:
-      #   predict(x, y) -> Float
+      #   score(x, y) -> Float
       #
       # * *Arguments* :
       #   - +x+ (NMatrix, shape: [n_samples, n_features]) -- Testing data.

data/lib/svmkit/version.rb

@@ -1,3 +1,3 @@
 module SVMKit
-  VERSION = '0.1.0'.freeze
+  VERSION = '0.1.1'.freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: svmkit
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors:
 - yoshoku
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2017-09-30 00:00:00.000000000 Z
+date: 2017-10-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -93,6 +93,7 @@ files:
 - lib/svmkit/base/classifier.rb
 - lib/svmkit/base/transformer.rb
 - lib/svmkit/kernel_approximation/rbf.rb
+- lib/svmkit/linear_model/logistic_regression.rb
 - lib/svmkit/linear_model/pegasos_svc.rb
 - lib/svmkit/multiclass/one_vs_rest_classifier.rb
 - lib/svmkit/preprocessing/l2_normalizer.rb