lightgbm 0.1.4 → 0.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 81f1f695112234bb576afaab35f4bf276d1f9c4a4adf0c74831cd1bb73f6baa0
-  data.tar.gz: 59ef1f3c581f83e108ce2a6f2c847bb7488fc2ea7f39ba217aaeffbf46e99351
+  metadata.gz: dad09a50916e229119d99de66aa676868224af44d8958916284718edb7fb3a3f
+  data.tar.gz: 2f7d299a7037455a4ebd1ccaf41bc102ec66808b9e806a59194fd14699204364
 SHA512:
-  metadata.gz: 21297d26e88957dd60d0aa61da19aa53aa632958adaf069b0efe2c8dae35e2e21e74c374da3509e337ca3268613b14dc541aee5012df086af7e8f784adb5063d
-  data.tar.gz: 7dbdc0fccaf256a1a835aea3eaa51fe326a0cd4b8cde168a6b8c27ff00c6412a5b5d82583fb1c14749f001f002a4fa3c3e156b9c4b2174b1e91f9449a4fa9ba1
+  metadata.gz: aa415946a82be07f1b9b8823b60fc75030f9e5547859078459583fa64d9212aad11db39d30e8e327b1fe23f48fc2ce9c6ddbd0445b52f73b44374a07c4724f5c
+  data.tar.gz: 99357888fde412672ecd804396cbd3a524b7dee6318d46d92a07064b858cd5f62f144f6b680a25ae0f1ae567df7d4670fcf38ba706fa31cbdfe58fd79eabc62d

data/CHANGELOG.md

@@ -1,3 +1,10 @@
+## 0.1.5
+
+- Packaged LightGBM with gem
+- Added support for missing values
+- Added `feature_names` to datasets
+- Fixed Daru training and prediction
+
 ## 0.1.4
 
 - Friendlier message when LightGBM not found

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

@@ -8,19 +8,23 @@
 
 ## 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'
 ```
 
+LightGBM is packaged with the gem, so no need to install it separately. On Mac, also run:
+
+```sh
+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_` prefix is removed from methods
+- The `get_` and `set_` prefixes are removed from methods
 - The default verbosity is `-1`
 - With the `cv` method, `stratified` is set to `false`
 
@@ -28,18 +32,25 @@ Some methods and options are also missing at the moment. PRs welcome!
 
 ## Training API
 
+Prep your data
+
+```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

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"
@@ -39,6 +40,8 @@ module LightGBM
         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

data/lib/lightgbm/booster.rb

@@ -116,7 +116,12 @@ module LightGBM
 
     # 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
@@ -124,8 +129,10 @@ module LightGBM
       num_iteration ||= best_iteration
       num_class ||= num_class()
 
+      flat_input = input.flatten
+      handle_missing(flat_input)
       data = ::FFI::MemoryPointer.new(:float, input.count * input.first.count)
-      data.put_array_of_float(0, input.flatten)
+      data.put_array_of_float(0, flat_input)
 
       out_len = ::FFI::MemoryPointer.new(:int64)
       out_result = ::FFI::MemoryPointer.new(:double, num_class * input.count)

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

@@ -2,12 +2,14 @@ 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
 
       # TODO stringify params
       params ||= {}
-      params["categorical_feature"] ||= categorical_feature.join(",") if categorical_feature != "auto"
+      if categorical_feature != "auto" && categorical_feature.any?
+        params["categorical_feature"] ||= categorical_feature.join(",")
+      end
       set_verbosity(params)
 
       @handle = ::FFI::MemoryPointer.new(:pointer)
@@ -26,7 +28,7 @@ module LightGBM
           flat_data = data.to_a.flatten
         elsif daru?(data)
           nrow, ncol = data.shape
-          flat_data = data.each_vector.map(&:to_a).flatten
+          flat_data = data.map_rows(&:to_a).flatten
         elsif narray?(data)
           nrow, ncol = data.shape
           flat_data = data.flatten.to_a
@@ -36,6 +38,7 @@ module LightGBM
           flat_data = data.flatten
         end
 
+        handle_missing(flat_data)
         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)
@@ -45,6 +48,7 @@ module LightGBM
       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 label
@@ -59,6 +63,16 @@ module LightGBM
       set_field("label", label)
     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(:string, 255) }
+      out_strs.put_array_of_pointer(0, 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 weight=(weight)
       set_field("weight", weight)
     end
@@ -67,6 +81,12 @@ module LightGBM
       set_field("group", group, type: :int32)
     end
 
+    def 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
+
     def num_data
       out = ::FFI::MemoryPointer.new(:int)
       check_result FFI.LGBM_DatasetGetNumData(handle_pointer, out)
@@ -131,18 +151,6 @@ module LightGBM
       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

@@ -6,6 +6,9 @@ module LightGBM
       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"
+      end
       raise LoadError, "Could not find LightGBM"
     end
 
@@ -19,6 +22,8 @@ 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

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 +23,22 @@ 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 narray?(data)
+      defined?(Numo::NArray) && data.is_a?(Numo::NArray)
+    end
   end
 end

data/lib/lightgbm/version.rb

@@ -1,3 +1,3 @@
 module LightGBM
-  VERSION = "0.1.4"
+  VERSION = "0.1.5"
 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: 0.1.5
 platform: ruby
 authors:
 - Andrew Kane
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-08-19 00:00:00.000000000 Z
+date: 2019-09-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -101,6 +101,7 @@ extensions: []
 extra_rdoc_files: []
 files:
 - CHANGELOG.md
+- LICENSE.txt
 - README.md
 - lib/lightgbm.rb
 - lib/lightgbm/booster.rb
@@ -112,6 +113,10 @@ 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