DecisionTree: Gini index split criteria added (#215)

Author: gyrdym

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## 16.8.0
+- DecisionTreeClassifier:
+    - Added Gini index assessor type
+
 ## 16.7.2
 - DecisionTreeClassifier:
     - TreeNode fields renamed

e2e/decision_tree_classifier/iris_tree.svg

@@ -14,3 +14,4 @@ export 'package:ml_algo/src/regressor/knn_regressor/knn_regressor.dart';
 export 'package:ml_algo/src/regressor/linear_regressor/linear_regressor.dart';
 export 'package:ml_algo/src/retrieval/kd_tree/kd_tree.dart';
 export 'package:ml_algo/src/retrieval/kd_tree/kd_tree_split_strategy.dart';
+export 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_type.dart';

e2e/decision_tree_classifier/pima_indians_tree.svg

@@ -2,16 +2,16 @@ import 'package:injector/injector.dart';
 import 'package:ml_algo/src/classifier/decision_tree_classifier/_injector.dart';
 import 'package:ml_algo/src/classifier/decision_tree_classifier/decision_tree_classifier_factory.dart';
 import 'package:ml_algo/src/classifier/decision_tree_classifier/decision_tree_classifier_factory_impl.dart';
+import 'package:ml_algo/src/common/distribution_calculator/distribution_calculator.dart';
 import 'package:ml_algo/src/common/distribution_calculator/distribution_calculator_factory.dart';
 import 'package:ml_algo/src/common/distribution_calculator/distribution_calculator_factory_impl.dart';
+import 'package:ml_algo/src/common/distribution_calculator/distribution_calculator_impl.dart';
 import 'package:ml_algo/src/di/common/init_common_module.dart';
 import 'package:ml_algo/src/extensions/injector.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_detector/leaf_detector_factory.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_detector/leaf_detector_factory_impl.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_label/leaf_label_factory_factory.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_label/leaf_label_factory_factory_impl.dart';
-import 'package:ml_algo/src/tree_trainer/split_assessor/split_assessor_factory.dart';
-import 'package:ml_algo/src/tree_trainer/split_assessor/split_assessor_factory_impl.dart';
 import 'package:ml_algo/src/tree_trainer/split_selector/split_selector_factory.dart';
 import 'package:ml_algo/src/tree_trainer/split_selector/split_selector_factory_impl.dart';
 import 'package:ml_algo/src/tree_trainer/splitter/nominal_splitter/nominal_splitter_factory.dart';
@@ -20,6 +20,8 @@ import 'package:ml_algo/src/tree_trainer/splitter/numerical_splitter/numerical_s
 import 'package:ml_algo/src/tree_trainer/splitter/numerical_splitter/numerical_splitter_factory_impl.dart';
 import 'package:ml_algo/src/tree_trainer/splitter/splitter_factory.dart';
 import 'package:ml_algo/src/tree_trainer/splitter/splitter_factory_impl.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_factory.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_factory_impl.dart';
 import 'package:ml_algo/src/tree_trainer/tree_trainer_factory.dart';
 import 'package:ml_algo/src/tree_trainer/tree_trainer_factory_impl.dart';
 
@@ -29,25 +31,27 @@ Injector initDecisionTreeModule() {
   return decisionTreeInjector
     ..registerSingletonIf<DistributionCalculatorFactory>(
         () => const DistributionCalculatorFactoryImpl())
+    ..registerSingletonIf<DistributionCalculator>(
+        () => const DistributionCalculatorImpl())
     ..registerSingletonIf<NominalTreeSplitterFactory>(
         () => const NominalTreeSplitterFactoryImpl())
     ..registerSingletonIf<NumericalTreeSplitterFactory>(
         () => const NumericalTreeSplitterFactoryImpl())
-    ..registerSingletonIf<TreeSplitAssessorFactory>(
-        () => const TreeSplitAssessorFactoryImpl())
+    ..registerSingletonIf<TreeAssessorFactory>(() => TreeAssessorFactoryImpl(
+        decisionTreeInjector.get<DistributionCalculator>()))
     ..registerSingletonIf<TreeSplitterFactory>(() => TreeSplitterFactoryImpl(
-          decisionTreeInjector.get<TreeSplitAssessorFactory>(),
+          decisionTreeInjector.get<TreeAssessorFactory>(),
           decisionTreeInjector.get<NominalTreeSplitterFactory>(),
           decisionTreeInjector.get<NumericalTreeSplitterFactory>(),
         ))
     ..registerSingletonIf<TreeSplitSelectorFactory>(
         () => TreeSplitSelectorFactoryImpl(
-              decisionTreeInjector.get<TreeSplitAssessorFactory>(),
+              decisionTreeInjector.get<TreeAssessorFactory>(),
               decisionTreeInjector.get<TreeSplitterFactory>(),
             ))
     ..registerSingletonIf<TreeLeafDetectorFactory>(
         () => TreeLeafDetectorFactoryImpl(
-              decisionTreeInjector.get<TreeSplitAssessorFactory>(),
+              decisionTreeInjector.get<TreeAssessorFactory>(),
             ))
     ..registerSingletonIf<TreeLeafLabelFactoryFactory>(
         () => TreeLeafLabelFactoryFactoryImpl(

lib/ml_algo.dart

@@ -7,13 +7,14 @@ import 'package:ml_algo/src/common/constants/default_parameters/common.dart';
 import 'package:ml_algo/src/common/serializable/serializable.dart';
 import 'package:ml_algo/src/model_selection/assessable.dart';
 import 'package:ml_algo/src/predictor/retrainable.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_type.dart';
 import 'package:ml_dataframe/ml_dataframe.dart';
 import 'package:ml_linalg/dtype.dart';
 
 /// A class that performs decision tree-based classification
 ///
 /// Decision tree is an algorithm that recursively splits the input data into
-/// subsets until the subsets conforming certain stop criteria are found.
+/// subsets until the subsets conforming certain stop criteria are met.
 ///
 /// Process of forming such a recursive subsets structure is called
 /// decision tree learning. Once a decision tree learned, it may be used to
@@ -45,13 +46,26 @@ abstract class DecisionTreeClassifier
   /// equal to [minSamplesCount] observations, the node turns into the leaf.
   ///
   /// [maxDepth] A maximum number of decision tree levels.
+  ///
+  /// [assessorType] Defines an assessment type that will be applied to a subset
+  /// of data in order to decide how to split the subset while building the tree.
+  /// Default value is [TreeAssessorType.gini]
+  ///
+  /// Possible values of [assessorType] :
+  ///
+  /// [TreeAssessorType.gini] The algorithm makes a decision on how to split a
+  /// subset of data based on the [Gini index](https://en.wikipedia.org/wiki/Gini_coefficient)
+  ///
+  /// [TreeAssessorType.majority] The algorithm makes a decision on how to split a
+  /// subset of data based on a major class.
   factory DecisionTreeClassifier(
     DataFrame trainData,
     String targetName, {
     num minError = 0.5,
     int minSamplesCount = 1,
     int maxDepth = 10,
     DType dtype = dTypeDefaultValue,
+    TreeAssessorType assessorType = TreeAssessorType.gini,
   }) =>
       initDecisionTreeModule().get<DecisionTreeClassifierFactory>().create(
             trainData,
@@ -60,6 +74,7 @@ abstract class DecisionTreeClassifier
             minError,
             minSamplesCount,
             maxDepth,
+            assessorType,
           );
 
   /// Restores previously fitted classifier instance from the given [json]
@@ -125,6 +140,10 @@ abstract class DecisionTreeClassifier
   /// The value is read-only, it's a hyperparameter of the model
   int get maxDepth;
 
+  /// An assessment type that was applied to a subset of data in order to
+  /// decide how to split the subset while building the tree
+  TreeAssessorType get assessorType;
+
   /// Saves tree as SVG-image. Example:
   ///
   /// ```dart

lib/src/classifier/decision_tree_classifier/_init_module.dart

@@ -1,4 +1,5 @@
-import 'package:ml_algo/ml_algo.dart';
+import 'package:ml_algo/src/classifier/decision_tree_classifier/decision_tree_classifier.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_type.dart';
 import 'package:ml_dataframe/ml_dataframe.dart';
 import 'package:ml_linalg/dtype.dart';
 
@@ -10,6 +11,7 @@ abstract class DecisionTreeClassifierFactory {
     num minError,
     int minSamplesCount,
     int maxDepth,
+    TreeAssessorType assessorType,
   );
 
   DecisionTreeClassifier fromJson(String json);

lib/src/classifier/decision_tree_classifier/decision_tree_classifier.dart

@@ -7,7 +7,7 @@ import 'package:ml_algo/src/helpers/validate_tree_solver_max_depth.dart';
 import 'package:ml_algo/src/helpers/validate_tree_solver_min_error.dart';
 import 'package:ml_algo/src/helpers/validate_tree_solver_min_samples_count.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_label/leaf_label_factory_type.dart';
-import 'package:ml_algo/src/tree_trainer/split_assessor/split_assessor_type.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_type.dart';
 import 'package:ml_algo/src/tree_trainer/split_selector/split_selector_type.dart';
 import 'package:ml_algo/src/tree_trainer/splitter/splitter_type.dart';
 import 'package:ml_algo/src/tree_trainer/tree_trainer_factory.dart';
@@ -29,6 +29,7 @@ class DecisionTreeClassifierFactoryImpl
     num minError,
     int minSamplesCount,
     int maxDepth,
+    TreeAssessorType assessorType,
   ) {
     validateTreeSolverMinError(minError);
     validateTreeSolversMinSamplesCount(minSamplesCount);
@@ -41,10 +42,10 @@ class DecisionTreeClassifierFactoryImpl
       minError,
       minSamplesCount,
       maxDepth,
-      TreeSplitAssessorType.majority,
+      assessorType,
       TreeLeafLabelFactoryType.majority,
       TreeSplitSelectorType.greedy,
-      TreeSplitAssessorType.majority,
+      assessorType,
       TreeSplitterType.greedy,
     );
     final treeRootNode = trainer.train(trainData.toMatrix(dtype));
@@ -55,6 +56,7 @@ class DecisionTreeClassifierFactoryImpl
       maxDepth,
       treeRootNode,
       targetName,
+      assessorType,
       dtype,
     );
   }

lib/src/classifier/decision_tree_classifier/decision_tree_classifier_factory.dart

@@ -12,6 +12,9 @@ import 'package:ml_algo/src/common/constants/common_json_keys.dart';
 import 'package:ml_algo/src/common/json_converter/dtype_json_converter.dart';
 import 'package:ml_algo/src/common/serializable/serializable_mixin.dart';
 import 'package:ml_algo/src/tree_trainer/leaf_label/leaf_label.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/helpers/from_tree_assessor_type_json.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/helpers/to_tree_assessor_type_json.dart';
+import 'package:ml_algo/src/tree_trainer/tree_assessor/tree_assessor_type.dart';
 import 'package:ml_algo/src/tree_trainer/tree_node/tree_node.dart';
 import 'package:ml_dataframe/ml_dataframe.dart';
 import 'package:ml_linalg/dtype.dart';
@@ -34,6 +37,7 @@ class DecisionTreeClassifierImpl
     this.maxDepth,
     this.treeRootNode,
     this.targetColumnName,
+    this.assessorType,
     this.dtype, {
     this.schemaVersion = decisionTreeClassifierJsonSchemaVersion,
   });
@@ -79,6 +83,13 @@ class DecisionTreeClassifierImpl
   @JsonKey(name: jsonSchemaVersionJsonKey)
   final int schemaVersion;
 
+  @override
+  @JsonKey(
+      name: decisionTreeClassifierAssessorTypeJsonKey,
+      toJson: toTreeAssessorTypeJson,
+      fromJson: fromTreeAssessorTypeJson)
+  final TreeAssessorType assessorType;
+
   @override
   DataFrame predict(DataFrame features) {
     final predictedLabels = features
@@ -144,6 +155,7 @@ class DecisionTreeClassifierImpl
           minError,
           minSamplesCount,
           maxDepth,
+          assessorType,
         );
   }