lightgbm 0.1.7 → 0.1.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cf13db584839dc0cb0a9b6d2480a6846704aee74f6a984300db06901cfbb1e66
-  data.tar.gz: ec716f3a2354c28e916b4ae89bd567e222f8bb5e5415e9579fb3c67def9c8edc
+  metadata.gz: d3b86686bc4575d069e469fb9c8911e93c7dcfb7622a415dc3e7ebee301947c3
+  data.tar.gz: a186cca11d4838fd13573b14ea154fff174804b31ae45236130c30a24f55ff7b
 SHA512:
-  metadata.gz: db7d6da179096a0648b2257c575383e9a59caa6bae04f39275aafe9126e69b51c6048c1452d2fb8dffa538362614efe83f7c05a8df4c749808040835a977623c
-  data.tar.gz: 5ebecf54455d3aef8ff3235e196bc064353bb7379a8b6fa61671feb223813ac6d028d7b86a0c9ada0568b06d048c6fcfdea48f082d208edaecbfdb54ead75d33
+  metadata.gz: 32ef8f452075bcf4441b8e0fcf68e63332c42da61445301ecd0a85b51b2af1b7a730da0810e1b182ff5fdfe184884242a5489469d16e0e9a717d19bb145eb095
+  data.tar.gz: 41b76c5ac174b75ce4e4f21477e4d32a9607b550e70448dde2797a765813d08a6c75de2f7f6d04ea79b4e9fa8afd768d097be516781e27e4eee4f982af364dc0

data/CHANGELOG.md

@@ -1,3 +1,8 @@
+## 0.1.8 (2020-05-09)
+
+- Improved error message when OpenMP not found on Mac
+- Fixed `Cannot add validation data` error
+
 ## 0.1.7 (2019-12-05)
 
 - Updated LightGBM to 2.3.1

data/README.md

@@ -1,8 +1,6 @@
 # LightGBM
 
-[LightGBM](https://github.com/microsoft/LightGBM) - the high performance machine learning library - for Ruby
-
-:fire: Uses the C API for blazing performance
+[LightGBM](https://github.com/microsoft/LightGBM) - high performance gradient boosting - for Ruby
 
 [![Build Status](https://travis-ci.org/ankane/lightgbm.svg?branch=master)](https://travis-ci.org/ankane/lightgbm)
 
@@ -20,16 +18,6 @@ On Mac, also install OpenMP:
 brew install libomp
 ```
 
-## Getting Started
-
-This library follows the [Python API](https://lightgbm.readthedocs.io/en/latest/Python-API.html). A few differences are:
-
-- The `get_` and `set_` prefixes are removed from methods
-- The default verbosity is `-1`
-- With the `cv` method, `stratified` is set to `false`
-
-Some methods and options are also missing at the moment. PRs welcome!
-
 ## Training API
 
 Prep your data
@@ -160,12 +148,18 @@ Numo::DFloat.new(3, 2).seq
 
 ## Related Projects
 
-- [Xgb](https://github.com/ankane/xgb) - XGBoost for Ruby
-- [Eps](https://github.com/ankane/eps) - Machine Learning for Ruby
+- [XGBoost](https://github.com/ankane/xgboost) - XGBoost for Ruby
+- [Eps](https://github.com/ankane/eps) - Machine learning for Ruby
 
 ## Credits
 
-Thanks to the [xgboost](https://github.com/PairOnAir/xgboost-ruby) gem for serving as an initial reference.
+This library follows the [Python API](https://lightgbm.readthedocs.io/en/latest/Python-API.html). A few differences are:
+
+- The `get_` and `set_` prefixes are removed from methods
+- The default verbosity is `-1`
+- With the `cv` method, `stratified` is set to `false`
+
+Thanks to the [xgboost](https://github.com/PairOnAir/xgboost-ruby) gem for showing how to use FFI.
 
 ## History
 
@@ -180,7 +174,7 @@ Everyone is encouraged to help improve this project. Here are a few ways you can
 - Write, clarify, or fix documentation
 - Suggest or add new features
 
-To get started with development and testing:
+To get started with development:
 
 ```sh
 git clone https://github.com/ankane/lightgbm.git

data/lib/lightgbm.rb

@@ -36,6 +36,8 @@ module LightGBM
           booster.train_data_name = name || "training"
           valid_contain_train = true
         else
+          # ensure the validation set references the training set
+          data.reference = train_set
           booster.add_valid(data, name || "valid_#{i}")
         end
       end
@@ -133,6 +135,7 @@ module LightGBM
       if early_stopping_rounds
         best_score = {}
         best_iter = {}
+        best_iteration = nil
       end
 
       num_boost_round.times do |iteration|
@@ -172,6 +175,7 @@ module LightGBM
               best_score[k] = score
               best_iter[k] = iteration
             elsif iteration - best_iter[k] >= early_stopping_rounds
+              best_iteration = best_iter[k]
               stop_early = true
               break
             end
@@ -180,6 +184,15 @@ module LightGBM
         end
       end
 
+      if early_stopping_rounds
+        # use best iteration from first metric if not stopped early
+        best_iteration ||= best_iter[best_iter.keys.first]
+        eval_hist.each_key do |k|
+          # TODO uncomment for 0.2.0
+          # eval_hist[k] = eval_hist[k].first(best_iteration + 1)
+        end
+      end
+
       eval_hist
     end
 

data/lib/lightgbm/dataset.rb

@@ -4,51 +4,16 @@ module LightGBM
 
     def initialize(data, label: nil, weight: nil, group: nil, params: nil, reference: nil, used_indices: nil, categorical_feature: "auto", feature_names: nil)
       @data = data
+      @label = label
+      @weight = weight
+      @group = group
+      @params = params
+      @reference = reference
+      @used_indices = used_indices
+      @categorical_feature = categorical_feature
+      @feature_names = feature_names
 
-      # TODO stringify params
-      params ||= {}
-      if categorical_feature != "auto" && categorical_feature.any?
-        params["categorical_feature"] ||= categorical_feature.join(",")
-      end
-      set_verbosity(params)
-
-      @handle = ::FFI::MemoryPointer.new(:pointer)
-      parameters = params_str(params)
-      reference = reference.handle_pointer if reference
-      if used_indices
-        used_row_indices = ::FFI::MemoryPointer.new(:int32, used_indices.count)
-        used_row_indices.write_array_of_int32(used_indices)
-        check_result FFI.LGBM_DatasetGetSubset(reference, used_row_indices, used_indices.count, parameters, @handle)
-      elsif data.is_a?(String)
-        check_result FFI.LGBM_DatasetCreateFromFile(data, parameters, reference, @handle)
-      else
-        if matrix?(data)
-          nrow = data.row_count
-          ncol = data.column_count
-          flat_data = data.to_a.flatten
-        elsif daru?(data)
-          nrow, ncol = data.shape
-          flat_data = data.map_rows(&:to_a).flatten
-        elsif narray?(data)
-          nrow, ncol = data.shape
-          flat_data = data.flatten.to_a
-        else
-          nrow = data.count
-          ncol = data.first.count
-          flat_data = data.flatten
-        end
-
-        handle_missing(flat_data)
-        c_data = ::FFI::MemoryPointer.new(:double, nrow * ncol)
-        c_data.write_array_of_double(flat_data)
-        check_result FFI.LGBM_DatasetCreateFromMat(c_data, 1, nrow, ncol, 1, parameters, reference, @handle)
-      end
-      ObjectSpace.define_finalizer(self, self.class.finalize(handle_pointer)) unless used_indices
-
-      self.label = label if label
-      self.weight = weight if weight
-      self.group = group if group
-      self.feature_names = feature_names if feature_names
+      construct
     end
 
     def label
@@ -59,10 +24,6 @@ module LightGBM
       field("weight")
     end
 
-    def label=(label)
-      set_field("label", label)
-    end
-
     def feature_names
       # must preallocate space
       num_feature_names = ::FFI::MemoryPointer.new(:int)
@@ -73,20 +34,37 @@ module LightGBM
       str_ptrs[0, num_feature_names.read_int].map(&:read_string)
     end
 
+    def label=(label)
+      @label = label
+      set_field("label", label)
+    end
+
     def weight=(weight)
+      @weight = weight
       set_field("weight", weight)
     end
 
     def group=(group)
+      @group = group
       set_field("group", group, type: :int32)
     end
 
     def feature_names=(feature_names)
+      @feature_names = feature_names
       c_feature_names = ::FFI::MemoryPointer.new(:pointer, feature_names.size)
       c_feature_names.write_array_of_pointer(feature_names.map { |v| ::FFI::MemoryPointer.from_string(v) })
       check_result FFI.LGBM_DatasetSetFeatureNames(handle_pointer, c_feature_names, feature_names.size)
     end
 
+    # TODO only update reference if not in chain
+    def reference=(reference)
+      if reference != @reference
+        @reference = reference
+        free_handle
+        construct
+      end
+    end
+
     def num_data
       out = ::FFI::MemoryPointer.new(:int)
       check_result FFI.LGBM_DatasetGetNumData(handle_pointer, out)
@@ -124,6 +102,61 @@ module LightGBM
 
     private
 
+    def construct
+      data = @data
+      used_indices = @used_indices
+
+      # TODO stringify params
+      params = @params || {}
+      if @categorical_feature != "auto" && @categorical_feature.any?
+        params["categorical_feature"] ||= @categorical_feature.join(",")
+      end
+      set_verbosity(params)
+
+      @handle = ::FFI::MemoryPointer.new(:pointer)
+      parameters = params_str(params)
+      reference = @reference.handle_pointer if @reference
+      if used_indices
+        used_row_indices = ::FFI::MemoryPointer.new(:int32, used_indices.count)
+        used_row_indices.write_array_of_int32(used_indices)
+        check_result FFI.LGBM_DatasetGetSubset(reference, used_row_indices, used_indices.count, parameters, @handle)
+      elsif data.is_a?(String)
+        check_result FFI.LGBM_DatasetCreateFromFile(data, parameters, reference, @handle)
+      else
+        if matrix?(data)
+          nrow = data.row_count
+          ncol = data.column_count
+          flat_data = data.to_a.flatten
+        elsif daru?(data)
+          nrow, ncol = data.shape
+          flat_data = data.map_rows(&:to_a).flatten
+        elsif narray?(data)
+          nrow, ncol = data.shape
+          flat_data = data.flatten.to_a
+        else
+          nrow = data.count
+          ncol = data.first.count
+          flat_data = data.flatten
+        end
+
+        handle_missing(flat_data)
+        c_data = ::FFI::MemoryPointer.new(:double, nrow * ncol)
+        c_data.write_array_of_double(flat_data)
+        check_result FFI.LGBM_DatasetCreateFromMat(c_data, 1, nrow, ncol, 1, parameters, reference, @handle)
+      end
+      ObjectSpace.define_finalizer(self, self.class.finalize(handle_pointer)) unless used_indices
+
+      self.label = @label if @label
+      self.weight = @weight if @weight
+      self.group = @group if @group
+      self.feature_names = @feature_names if @feature_names
+    end
+
+    def free_handle
+      FFI.LGBM_DatasetFree(handle_pointer)
+      ObjectSpace.undefine_finalizer(self)
+    end
+
     def dump_text(filename)
       check_result FFI.LGBM_DatasetDumpText(handle_pointer, filename)
     end

data/lib/lightgbm/ffi.rb

@@ -5,11 +5,11 @@ module LightGBM
     begin
       ffi_lib LightGBM.ffi_lib
     rescue LoadError => e
-      raise e if ENV["LIGHTGBM_DEBUG"]
-      if e.message.include?("libomp")
-        raise LoadError, "Could not find OpenMP"
+      if e.message.include?("Library not loaded: /usr/local/opt/libomp/lib/libomp.dylib") && e.message.include?("Reason: image not found")
+        raise LoadError, "OpenMP not found. Run `brew install libomp`"
+      else
+        raise e
       end
-      raise LoadError, "Could not find LightGBM"
     end
 
     # https://github.com/microsoft/LightGBM/blob/master/include/LightGBM/c_api.h

data/lib/lightgbm/version.rb

@@ -1,3 +1,3 @@
 module LightGBM
-  VERSION = "0.1.7"
+  VERSION = "0.1.8"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: lightgbm
 version: !ruby/object:Gem::Version
-  version: 0.1.7
+  version: 0.1.8
 platform: ruby
 authors:
 - Andrew Kane
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-12-06 00:00:00.000000000 Z
+date: 2020-05-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -122,8 +122,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.3
+rubygems_version: 3.1.2
 signing_key: 
 specification_version: 4
-summary: LightGBM - the high performance machine learning library - for Ruby
+summary: High performance gradient boosting for Ruby
 test_files: []