lightgbm 0.1.4 → 0.1.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 81f1f695112234bb576afaab35f4bf276d1f9c4a4adf0c74831cd1bb73f6baa0
-  data.tar.gz: 59ef1f3c581f83e108ce2a6f2c847bb7488fc2ea7f39ba217aaeffbf46e99351
+  metadata.gz: a4aac9eac1ab0dadbf31d1e1fc2714e75a8c37075538aef5c53b42df1c34f658
+  data.tar.gz: ca3a1043c55184992b3fac611963062d01747449f6170719ef1f299d4f0474c9
 SHA512:
-  metadata.gz: 21297d26e88957dd60d0aa61da19aa53aa632958adaf069b0efe2c8dae35e2e21e74c374da3509e337ca3268613b14dc541aee5012df086af7e8f784adb5063d
-  data.tar.gz: 7dbdc0fccaf256a1a835aea3eaa51fe326a0cd4b8cde168a6b8c27ff00c6412a5b5d82583fb1c14749f001f002a4fa3c3e156b9c4b2174b1e91f9449a4fa9ba1
+  metadata.gz: c2f14ccc3b40690060d2ee533cfe46e137f40e91520aab3eb188a8a03a697a6956853b717e743cd357d0a059836fc32e9e0aa0fbe5a7dd4263b1ff3e94e79601
+  data.tar.gz: 64abdb43f4c45222dcbb39ddde1d21350b2b4958439dc84d8f13684f4e9bcd3aa2c0e49aa8fd9ae1dfc09b8bc0ce69592e95aab0542684f3c952e9e67ac4689f

data/CHANGELOG.md

@@ -1,4 +1,31 @@
-## 0.1.4
+## 0.1.9 (2020-06-10)
+
+- Added support for Rover
+- Improved performance of Numo datasets
+
+## 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
+- Switched to doubles for datasets and predictions
+
+## 0.1.6 (2019-09-29)
+
+- Updated LightGBM to 2.3.0
+- Fixed error with JRuby
+
+## 0.1.5 (2019-09-03)
+
+- Packaged LightGBM with gem
+- Added support for missing values
+- Added `feature_names` to datasets
+- Fixed Daru training and prediction
+
+## 0.1.4 (2019-08-19)
 
 - Friendlier message when LightGBM not found
 - Added `Ranker`
@@ -6,22 +33,22 @@
 - Free memory when objects are destroyed
 - Removed unreleased `dump_text` method
 
-## 0.1.3
+## 0.1.3 (2019-08-16)
 
 - Added Scikit-Learn API
 - Added support for Daru and Numo::NArray
 
-## 0.1.2
+## 0.1.2 (2019-08-15)
 
 - Added `cv` method
 - Added early stopping
 - Fixed multiclass classification
 
-## 0.1.1
+## 0.1.1 (2019-08-14)
 
 - Added training API
 - Added many methods
 
-## 0.1.0
+## 0.1.0 (2019-08-13)
 
 - 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

@@ -1,45 +1,44 @@
 # 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)
 
 ## Installation
 
-First, [install LightGBM](https://lightgbm.readthedocs.io/en/latest/Installation-Guide.html). On Mac, copy `lib_lightgbm.so` to `/usr/local/lib`.
-
 Add this line to your application’s Gemfile:
 
 ```ruby
 gem 'lightgbm'
 ```
 
-## Getting Started
+On Mac, also install OpenMP:
 
-This library follows the [Python API](https://lightgbm.readthedocs.io/en/latest/Python-API.html). A few differences are:
+```sh
+brew install libomp
+```
 
-- The `get_` prefix is removed from methods
-- The default verbosity is `-1`
-- With the `cv` method, `stratified` is set to `false`
+## Training API
 
-Some methods and options are also missing at the moment. PRs welcome!
+Prep your data
 
-## Training API
+```ruby
+x = [[1, 2], [3, 4], [5, 6], [7, 8]]
+y = [1, 2, 3, 4]
+```
 
 Train a model
 
 ```ruby
 params = {objective: "regression"}
-train_set = LightGBM::Dataset.new(x_train, label: y_train)
+train_set = LightGBM::Dataset.new(x, label: y)
 booster = LightGBM.train(params, train_set)
 ```
 
 Predict
 
 ```ruby
-booster.predict(x_test)
+booster.predict(x)
 ```
 
 Save the model to a file
@@ -130,16 +129,22 @@ Data can be an array of arrays
 [[1, 2, 3], [4, 5, 6]]
 ```
 
-Or a Daru data frame
+Or a Numo NArray
 
 ```ruby
-Daru::DataFrame.from_csv("houses.csv")
+Numo::NArray.cast([[1, 2, 3], [4, 5, 6]])
 ```
 
-Or a Numo NArray
+Or a Rover data frame
+
+```ruby
+Rover.read_csv("houses.csv")
+```
+
+Or a Daru data frame
 
 ```ruby
-Numo::DFloat.new(3, 2).seq
+Daru::DataFrame.from_csv("houses.csv")
 ```
 
 ## Helpful Resources
@@ -149,12 +154,18 @@ 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, and Selva Prabhakaran for the [test datasets](https://github.com/selva86/datasets).
+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
 
@@ -168,3 +179,13 @@ Everyone is encouraged to help improve this project. Here are a few ways you can
 - Fix bugs and [submit pull requests](https://github.com/ankane/lightgbm/pulls)
 - Write, clarify, or fix documentation
 - Suggest or add new features
+
+To get started with development:
+
+```sh
+git clone https://github.com/ankane/lightgbm.git
+cd lightgbm
+bundle install
+bundle exec rake vendor:all
+bundle exec rake test
+```

data/lib/lightgbm.rb

@@ -20,7 +20,8 @@ module LightGBM
     attr_accessor :ffi_lib
   end
   lib_name = "lib_lightgbm.#{::FFI::Platform::LIBSUFFIX}"
-  self.ffi_lib = [lib_name, "lib_lightgbm.so"]
+  vendor_lib = File.expand_path("../vendor/#{lib_name}", __dir__)
+  self.ffi_lib = [lib_name, "lib_lightgbm.so", vendor_lib]
 
   # friendlier error message
   autoload :FFI, "lightgbm/ffi"
@@ -35,10 +36,14 @@ 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
 
+      raise ArgumentError, "For early stopping, at least one validation set is required" if early_stopping_rounds && !valid_sets.any? { |v| v != train_set }
+
       booster.best_iteration = 0
 
       if early_stopping_rounds
@@ -130,6 +135,7 @@ module LightGBM
       if early_stopping_rounds
         best_score = {}
         best_iter = {}
+        best_iteration = nil
       end
 
       num_boost_round.times do |iteration|
@@ -169,6 +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
@@ -177,6 +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/booster.rb

@@ -30,7 +30,7 @@ module LightGBM
 
     def current_iteration
       out = ::FFI::MemoryPointer.new(:int)
-      check_result FFI::LGBM_BoosterGetCurrentIteration(handle_pointer, out)
+      check_result FFI.LGBM_BoosterGetCurrentIteration(handle_pointer, out)
       out.read_int
     end
 
@@ -38,11 +38,11 @@ module LightGBM
       num_iteration ||= best_iteration
       buffer_len = 1 << 20
       out_len = ::FFI::MemoryPointer.new(:int64)
-      out_str = ::FFI::MemoryPointer.new(:string, buffer_len)
+      out_str = ::FFI::MemoryPointer.new(:char, buffer_len)
       check_result FFI.LGBM_BoosterDumpModel(handle_pointer, start_iteration, num_iteration, buffer_len, out_len, out_str)
-      actual_len = out_len.read_int64
+      actual_len = read_int64(out_len)
       if actual_len > buffer_len
-        out_str = ::FFI::MemoryPointer.new(:string, actual_len)
+        out_str = ::FFI::MemoryPointer.new(:char, actual_len)
         check_result FFI.LGBM_BoosterDumpModel(handle_pointer, start_iteration, num_iteration, actual_len, out_len, out_str)
       end
       out_str.read_string
@@ -85,11 +85,11 @@ module LightGBM
       num_iteration ||= best_iteration
       buffer_len = 1 << 20
       out_len = ::FFI::MemoryPointer.new(:int64)
-      out_str = ::FFI::MemoryPointer.new(:string, buffer_len)
+      out_str = ::FFI::MemoryPointer.new(:char, buffer_len)
       check_result FFI.LGBM_BoosterSaveModelToString(handle_pointer, start_iteration, num_iteration, buffer_len, out_len, out_str)
-      actual_len = out_len.read_int64
+      actual_len = read_int64(out_len)
       if actual_len > buffer_len
-        out_str = ::FFI::MemoryPointer.new(:string, actual_len)
+        out_str = ::FFI::MemoryPointer.new(:char, actual_len)
         check_result FFI.LGBM_BoosterSaveModelToString(handle_pointer, start_iteration, num_iteration, actual_len, out_len, out_str)
       end
       out_str.read_string
@@ -104,19 +104,24 @@ module LightGBM
 
     def num_model_per_iteration
       out = ::FFI::MemoryPointer.new(:int)
-      check_result FFI::LGBM_BoosterNumModelPerIteration(handle_pointer, out)
+      check_result FFI.LGBM_BoosterNumModelPerIteration(handle_pointer, out)
       out.read_int
     end
 
     def num_trees
       out = ::FFI::MemoryPointer.new(:int)
-      check_result FFI::LGBM_BoosterNumberOfTotalModel(handle_pointer, out)
+      check_result FFI.LGBM_BoosterNumberOfTotalModel(handle_pointer, out)
       out.read_int
     end
 
     # TODO support different prediction types
     def predict(input, num_iteration: nil, **params)
-      raise TypeError unless input.is_a?(Array)
+      input =
+        if daru?(input)
+          input.map_rows(&:to_a)
+        else
+          input.to_a
+        end
 
       singular = !input.first.is_a?(Array)
       input = [input] if singular
@@ -124,13 +129,15 @@ module LightGBM
       num_iteration ||= best_iteration
       num_class ||= num_class()
 
-      data = ::FFI::MemoryPointer.new(:float, input.count * input.first.count)
-      data.put_array_of_float(0, input.flatten)
+      flat_input = input.flatten
+      handle_missing(flat_input)
+      data = ::FFI::MemoryPointer.new(:double, input.count * input.first.count)
+      data.write_array_of_double(flat_input)
 
       out_len = ::FFI::MemoryPointer.new(:int64)
       out_result = ::FFI::MemoryPointer.new(:double, num_class * input.count)
-      check_result FFI.LGBM_BoosterPredictForMat(handle_pointer, data, 0, input.count, input.first.count, 1, 0, num_iteration, params_str(params), out_len, out_result)
-      out = out_result.read_array_of_double(out_len.read_int64)
+      check_result FFI.LGBM_BoosterPredictForMat(handle_pointer, data, 1, input.count, input.first.count, 1, 0, num_iteration, params_str(params), out_len, out_result)
+      out = out_result.read_array_of_double(read_int64(out_len))
       out = out.each_slice(num_class).to_a if num_class > 1
 
       singular ? out.first : out
@@ -161,7 +168,7 @@ module LightGBM
 
     def eval_counts
       out = ::FFI::MemoryPointer.new(:int)
-      check_result FFI::LGBM_BoosterGetEvalCounts(handle_pointer, out)
+      check_result FFI.LGBM_BoosterGetEvalCounts(handle_pointer, out)
       out.read_int
     end
 
@@ -169,8 +176,8 @@ module LightGBM
       eval_counts ||= eval_counts()
       out_len = ::FFI::MemoryPointer.new(:int)
       out_strs = ::FFI::MemoryPointer.new(:pointer, eval_counts)
-      str_ptrs = eval_counts.times.map { ::FFI::MemoryPointer.new(:string, 255) }
-      out_strs.put_array_of_pointer(0, str_ptrs)
+      str_ptrs = eval_counts.times.map { ::FFI::MemoryPointer.new(:char, 255) }
+      out_strs.write_array_of_pointer(str_ptrs)
       check_result FFI.LGBM_BoosterGetEvalNames(handle_pointer, out_len, out_strs)
       str_ptrs.map(&:read_string)
     end
@@ -191,10 +198,15 @@ module LightGBM
 
     def num_class
       out = ::FFI::MemoryPointer.new(:int)
-      check_result FFI::LGBM_BoosterGetNumClasses(handle_pointer, out)
+      check_result FFI.LGBM_BoosterGetNumClasses(handle_pointer, out)
       out.read_int
     end
 
+    # read_int64 not available on JRuby
+    def read_int64(ptr)
+      ptr.read_array_of_int64(1).first
+    end
+
     include Utils
   end
 end

data/lib/lightgbm/classifier.rb

@@ -15,8 +15,8 @@ module LightGBM
         params[:objective] ||= "binary"
       end
 
-      train_set = Dataset.new(x, label: y, categorical_feature: categorical_feature)
-      valid_sets = Array(eval_set).map { |v| Dataset.new(v[0], label: v[1], reference: train_set) }
+      train_set = Dataset.new(x, label: y, categorical_feature: categorical_feature, params: params)
+      valid_sets = Array(eval_set).map { |v| Dataset.new(v[0], label: v[1], reference: train_set, params: params) }
 
       @booster = LightGBM.train(params, train_set,
         num_boost_round: @n_estimators,

data/lib/lightgbm/dataset.rb

@@ -2,49 +2,18 @@ module LightGBM
   class Dataset
     attr_reader :data, :params
 
-    def initialize(data, label: nil, weight: nil, group: nil, params: nil, reference: nil, used_indices: nil, categorical_feature: "auto")
+    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 ||= {}
-      params["categorical_feature"] ||= categorical_feature.join(",") if categorical_feature != "auto"
-      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.put_array_of_int32(0, 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.each_vector.map(&: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
-
-        c_data = ::FFI::MemoryPointer.new(:float, nrow * ncol)
-        c_data.put_array_of_float(0, flat_data)
-        check_result FFI.LGBM_DatasetCreateFromMat(c_data, 0, 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
+      construct
     end
 
     def label
@@ -55,18 +24,47 @@ module LightGBM
       field("weight")
     end
 
+    def feature_names
+      # must preallocate space
+      num_feature_names = ::FFI::MemoryPointer.new(:int)
+      out_strs = ::FFI::MemoryPointer.new(:pointer, 1000)
+      str_ptrs = 1000.times.map { ::FFI::MemoryPointer.new(:char, 255) }
+      out_strs.write_array_of_pointer(str_ptrs)
+      check_result FFI.LGBM_DatasetGetFeatureNames(handle_pointer, out_strs, num_feature_names)
+      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)
@@ -83,11 +81,6 @@ module LightGBM
       check_result FFI.LGBM_DatasetSaveBinary(handle_pointer, filename)
     end
 
-    # not released yet
-    # def dump_text(filename)
-    #   check_result FFI.LGBM_DatasetDumpText(handle_pointer, filename)
-    # end
-
     def subset(used_indices, params: nil)
       # categorical_feature passed via params
       params ||= self.params
@@ -109,6 +102,70 @@ 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 numo?(data) || rover?(data)
+          data = data.to_numo if rover?(data)
+          nrow, ncol = data.shape
+        else
+          nrow = data.count
+          ncol = data.first.count
+          flat_data = data.flatten
+        end
+
+        c_data = ::FFI::MemoryPointer.new(:double, nrow * ncol)
+        if numo?(data)
+          c_data.write_bytes(data.cast_to(Numo::DFloat).to_string)
+        else
+          handle_missing(flat_data)
+          c_data.write_array_of_double(flat_data)
+        end
+
+        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
+
     def field(field_name)
       num_data = self.num_data
       out_len = ::FFI::MemoryPointer.new(:int)
@@ -122,27 +179,15 @@ module LightGBM
       data = data.to_a unless data.is_a?(Array)
       if type == :int32
         c_data = ::FFI::MemoryPointer.new(:int32, data.count)
-        c_data.put_array_of_int32(0, data)
+        c_data.write_array_of_int32(data)
         check_result FFI.LGBM_DatasetSetField(handle_pointer, field_name, c_data, data.count, 2)
       else
         c_data = ::FFI::MemoryPointer.new(:float, data.count)
-        c_data.put_array_of_float(0, data)
+        c_data.write_array_of_float(data)
         check_result FFI.LGBM_DatasetSetField(handle_pointer, field_name, c_data, data.count, 0)
       end
     end
 
-    def matrix?(data)
-      defined?(Matrix) && data.is_a?(Matrix)
-    end
-
-    def daru?(data)
-      defined?(Daru::DataFrame) && data.is_a?(Daru::DataFrame)
-    end
-
-    def narray?(data)
-      defined?(Numo::NArray) && data.is_a?(Numo::NArray)
-    end
-
     include Utils
   end
 end

data/lib/lightgbm/ffi.rb

@@ -5,8 +5,11 @@ module LightGBM
     begin
       ffi_lib LightGBM.ffi_lib
     rescue LoadError => e
-      raise e if ENV["LIGHTGBM_DEBUG"]
-      raise LoadError, "Could not find LightGBM"
+      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
     end
 
     # https://github.com/microsoft/LightGBM/blob/master/include/LightGBM/c_api.h
@@ -19,9 +22,11 @@ module LightGBM
     attach_function :LGBM_DatasetCreateFromFile, %i[string string pointer pointer], :int
     attach_function :LGBM_DatasetCreateFromMat, %i[pointer int int32 int32 int string pointer pointer], :int
     attach_function :LGBM_DatasetGetSubset, %i[pointer pointer int32 string pointer], :int
+    attach_function :LGBM_DatasetSetFeatureNames, %i[pointer pointer int], :int
+    attach_function :LGBM_DatasetGetFeatureNames, %i[pointer pointer pointer], :int
     attach_function :LGBM_DatasetFree, %i[pointer], :int
     attach_function :LGBM_DatasetSaveBinary, %i[pointer string], :int
-    # attach_function :LGBM_DatasetDumpText, %i[pointer string], :int
+    attach_function :LGBM_DatasetDumpText, %i[pointer string], :int
     attach_function :LGBM_DatasetSetField, %i[pointer string pointer int int], :int
     attach_function :LGBM_DatasetGetField, %i[pointer string pointer pointer pointer], :int
     attach_function :LGBM_DatasetGetNumData, %i[pointer pointer], :int

data/lib/lightgbm/regressor.rb

@@ -5,8 +5,8 @@ module LightGBM
     end
 
     def fit(x, y, categorical_feature: "auto", eval_set: nil, eval_names: [], early_stopping_rounds: nil, verbose: true)
-      train_set = Dataset.new(x, label: y, categorical_feature: categorical_feature)
-      valid_sets = Array(eval_set).map { |v| Dataset.new(v[0], label: v[1], reference: train_set) }
+      train_set = Dataset.new(x, label: y, categorical_feature: categorical_feature, params: @params)
+      valid_sets = Array(eval_set).map { |v| Dataset.new(v[0], label: v[1], reference: train_set, params: @params) }
 
       @booster = LightGBM.train(@params, train_set,
         num_boost_round: @n_estimators,

data/lib/lightgbm/utils.rb

@@ -23,5 +23,26 @@ module LightGBM
         params["verbosity"] = -1
       end
     end
+
+    # for categorical, NaN and negative value are the same
+    def handle_missing(data)
+      data.map! { |v| v.nil? ? Float::NAN : v }
+    end
+
+    def matrix?(data)
+      defined?(Matrix) && data.is_a?(Matrix)
+    end
+
+    def daru?(data)
+      defined?(Daru::DataFrame) && data.is_a?(Daru::DataFrame)
+    end
+
+    def numo?(data)
+      defined?(Numo::NArray) && data.is_a?(Numo::NArray)
+    end
+
+    def rover?(data)
+      defined?(Rover::DataFrame) && data.is_a?(Rover::DataFrame)
+    end
   end
 end

data/lib/lightgbm/version.rb

@@ -1,3 +1,3 @@
 module LightGBM
-  VERSION = "0.1.4"
+  VERSION = "0.1.9"
 end

data/vendor/LICENSE

@@ -0,0 +1,21 @@
+The MIT License (MIT)
+
+Copyright (c) Microsoft Corporation
+
+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.

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: lightgbm
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.9
 platform: ruby
 authors:
 - Andrew Kane
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-08-19 00:00:00.000000000 Z
+date: 2020-06-11 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -80,20 +80,6 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
-- !ruby/object:Gem::Dependency
-  name: numo-narray
-  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'
 description: 
 email: andrew@chartkick.com
 executables: []
@@ -101,6 +87,7 @@ extensions: []
 extra_rdoc_files: []
 files:
 - CHANGELOG.md
+- LICENSE.txt
 - README.md
 - lib/lightgbm.rb
 - lib/lightgbm/booster.rb
@@ -112,6 +99,10 @@ files:
 - lib/lightgbm/regressor.rb
 - lib/lightgbm/utils.rb
 - lib/lightgbm/version.rb
+- vendor/LICENSE
+- vendor/lib_lightgbm.dll
+- vendor/lib_lightgbm.dylib
+- vendor/lib_lightgbm.so
 homepage: https://github.com/ankane/lightgbm
 licenses:
 - MIT
@@ -131,8 +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: []