v1.0.16

- `ListComparablesExtension` changed to `IterableComparablesExtension`.
- New `IterableUint8ListExtension`, `ListAnswerExtension` and `ListOfListAnswer`.

CHANGELOG.md

@@ -1,3 +1,8 @@
+## 1.0.16
+
+- `ListComparablesExtension` changed to `IterableComparablesExtension`.
+- New `IterableUint8ListExtension`, `ListAnswerExtension` and `ListOfListAnswer`.
+
 ## 1.0.15
 
 - `Variable` node optimization:

lib/src/statistics_bayesnet.dart

@@ -8,8 +8,7 @@ Part of this code was inspired by the Java project:
 */
 
 import 'package:collection/collection.dart';
-
-import 'statistics_extension_num.dart';
+import 'package:statistics/statistics.dart';
 
 final _regexpSpace = RegExp(r'\s+');
 
@@ -202,7 +201,7 @@ class BayesianNetwork extends Iterable<String>
 
   void _checkNotFrozen() {
     if (isFrozen) {
-      throw StateError("Network already frozen!");
+      throw ValidationError("Network already frozen!");
     }
   }
 
@@ -221,17 +220,21 @@ class BayesianNetwork extends Iterable<String>
   }
 
   /// Returns `true` if contains a node with [name].
-  bool hasNode(String name) => _nodes.containsKey(name);
+  bool hasNodeWithName(String name) => _nodes.containsKey(name);
 
   /// Returns a [Variable] node with [name].
-  Variable getNode(String name) {
+  Variable getNodeByName(String name) {
     var node = _nodes[name];
     if (node == null) {
-      throw StateError("No such variable <" + name + ">.");
+      throw ValidationError("No `Variable` node with name: $name");
     }
     return node;
   }
 
+  /// Returns a [List] of [Variable]s nodes with matching [names].
+  List<Variable> getNodesByNames(Iterable<String> names) =>
+      _nodes.values.where((e) => names.contains(e.name)).toList();
+
   Condition _parseCondition(String line) {
     line = line.replaceAll(_regexpSpace, '');
     var cond = line.split(
@@ -483,7 +486,7 @@ class Variable extends Validatable implements Comparable<Variable> {
 
   @override
   Object? validate() {
-    if (_probabilities.values.any((n) => n.isNaN || n.isInfinite)) {
+    if (_probabilities.values.any((n) => n.isNaN || n.isInfinite || n < 0)) {
       return this;
     }
 
@@ -635,7 +638,7 @@ class Variable extends Validatable implements Comparable<Variable> {
     }
 
     var name = e[0];
-    var node = network.getNode(name);
+    var node = network.getNodeByName(name);
 
     return Event.byOutcomeName(node, e[1]);
   }
@@ -661,7 +664,7 @@ class Variable extends Validatable implements Comparable<Variable> {
     _checkNotFrozen();
 
     try {
-      _addParent(network.getNode(name));
+      _addParent(network.getNodeByName(name));
     } catch (e) {
       throw ValidationError(
           'The specified parent node "$name" does not exist (yet).');
@@ -753,13 +756,19 @@ class Variable extends Validatable implements Comparable<Variable> {
     return null;
   }
 
+  static final IterableEquality _iterableEquality = IterableEquality();
+
   @override
   bool operator ==(Object other) =>
       identical(this, other) ||
       other is Variable &&
           runtimeType == other.runtimeType &&
           name == other.name &&
-          network == other.network;
+          (identical(network, other.network) ||
+              (network.name == other.network.name &&
+                  network._nodes.length == other.network._nodes.length &&
+                  _iterableEquality.equals(
+                      network._nodes.keys, other.network._nodes.keys)));
 
   @override
   int get hashCode => name.hashCode ^ network.nodesLength;
@@ -1083,10 +1092,24 @@ class Answer implements Comparable<Answer> {
   /// The original unparsed query.
   final String originalQuery;
 
+  late final double fitness;
+
   Answer(this.query, this.targetValue, this.selectedValues, this.probability,
       {String? originalQuery, double? probabilityUnnormalized})
       : originalQuery = originalQuery ?? query,
-        probabilityUnnormalized = probabilityUnnormalized ?? probability;
+        probabilityUnnormalized = probabilityUnnormalized ?? probability,
+        fitness = _computeFitness(probability, probabilityUnnormalized);
+
+  static double _computeFitness(double probability,
+      [double? probabilityUnnormalized]) {
+    if (probabilityUnnormalized != null) {
+      var squaresSum = ((probability * probability) +
+          (probabilityUnnormalized * probabilityUnnormalized));
+      return squaresSum / 2;
+    } else {
+      return probability;
+    }
+  }
 
   @override
   bool operator ==(Object other) =>
@@ -1116,14 +1139,145 @@ class Answer implements Comparable<Answer> {
   int compareTo(Answer other) {
     if (identical(this, other)) return 0;
 
-    var cmp = probability.compareTo(other.probability);
+    var cmp = fitness.compareTo(other.fitness);
     if (cmp == 0) {
-      cmp = query.compareTo(other.query);
+      if (cmp == 0) {
+        cmp = probability.compareTo(other.probability);
+        if (cmp == 0) {
+          cmp = query.compareTo(other.query);
+        }
+      }
     }
     return cmp;
   }
 }
 
+extension ListAnswerExtension on List<Answer> {
+  /// Sorts the [Answer] [List] by [Answer.probability].
+  void sortByProbability() =>
+      sort((a, b) => a.probability.compareTo(b.probability));
+
+  /// Sorts the [Answer] [List] by [Answer.probabilityUnnormalized].
+  void sortByProbabilityUnnormalized() => sort(
+      (a, b) => a.probabilityUnnormalized.compareTo(b.probabilityUnnormalized));
+
+  /// Sorts the [Answer] [List] by [Answer.fitness].
+  void sortByFitness() => sort((a, b) => a.fitness.compareTo(b.fitness));
+
+  /// Sorts the [Answer] [List] by [Answer.selectedValues].
+  void sortBySelectedValues() {
+    sort((a, b) {
+      var cmp = a.selectedValues.compareWith(b.selectedValues);
+      if (cmp == 0) {
+        cmp = a.targetValue.compareTo(b.targetValue);
+      }
+      return cmp;
+    });
+  }
+
+  /// Sorts the [Answer] [List] by [Answer.selectedValues] signals.
+  void sortBySelectedValuesSignal() {
+    sort((a, b) {
+      var cmp = a.selectedValues
+          .map((e) => e.signal.index)
+          .compareWith(b.selectedValues.map((e) => e.signal.index));
+      if (cmp == 0) {
+        cmp = a.targetValue.compareTo(b.targetValue);
+      }
+      return cmp;
+    });
+  }
+
+  /// Sorts the [Answer] [List] by [Answer.targetValue].
+  void sortByTargetValue() {
+    sort((a, b) {
+      var cmp = a.targetValue.compareTo(b.targetValue);
+      if (cmp == 0) {
+        cmp = a.selectedValues.compareWith(b.selectedValues);
+      }
+      return cmp;
+    });
+  }
+
+  /// Sorts the [Answer] [List] by [Answer.targetValue].
+  void sortByQuery() => sort((a, b) => a.query.compareTo(b.query));
+
+  /// Groups [Answer]s by [Answer.selectedValues] [Variable].
+  Map<Variable, List<Answer>> groupBySelectedVariable() =>
+      this.groupBy((e) => e.selectedValues.first.variable);
+
+  /// Groups [Answer]s by [Answer.targetValue] [Variable].
+  Map<Variable, List<Answer>> groupByTargetVariable() =>
+      this.groupBy((e) => e.targetValue.variable);
+
+  /// Filters [Answer]s with [Answer.selectedValues] matching [signal].
+  List<Answer> withSelectedValueSignal(ValueSignal signal) =>
+      where((e) => e.selectedValues.any((e) => e.signal == signal)).toList();
+
+  /// Filters [Answer]s with [Answer.name] matching [valueName].
+  List<Answer> withSelectedValueName(String valueName) =>
+      where((e) => e.selectedValues.any((e) => e.name == valueName)).toList();
+
+  /// Returns the [Value] matching [valueName] [Answer.probability] mean.
+  double valueProbabilityMean(String valueName) =>
+      withSelectedValueName(valueName).map((e) => e.probability).mean;
+
+  /// Returns the [Value] matching [valueName] [Answer.probabilityUnnormalized] mean.
+  double valueProbabilityUnnormalizedMean(String valueName) =>
+      withSelectedValueName(valueName)
+          .map((e) => e.probabilityUnnormalized)
+          .mean;
+
+  /// Returns the performance by [Answer.probability] of opposite values:
+  ///
+  /// - Formula: `(positiveProbability + (1 - negativeProbability)) / 2`
+  ///   - positiveProbability: the mean probability of positive values.
+  ///   - negativeProbability: the mean probability of negative values.
+  double oppositeValuesPerformance(String positiveValue, String negativeValue) {
+    var positiveProb = valueProbabilityMean(positiveValue);
+    var negativeProb = valueProbabilityMean(negativeValue);
+    var performance = ((positiveProb) + (1 - negativeProb)) / 2;
+    return performance;
+  }
+
+  /// Returns the performance by [Answer.probabilityUnnormalized] of opposite values:
+  ///
+  /// - Formula: `(positiveProbability + (1 - negativeProbability)) / 2`
+  ///   - positiveProbability: the mean probability unnormalized of positive values.
+  ///   - negativeProbability: the mean probability unnormalized of negative values.
+  double oppositeValuesPerformanceUnnormalized(
+      String positiveValue, String negativeValue) {
+    var positiveProb = valueProbabilityUnnormalizedMean(positiveValue);
+    var negativeProb = valueProbabilityUnnormalizedMean(negativeValue);
+    var performance = ((positiveProb) + (1 - negativeProb)) / 2;
+    return performance;
+  }
+}
+
+extension ListOfListAnswerExtension on List<List<Answer>> {
+  /// Sort sub lists by [ListAnswerExtension.oppositeValuesPerformance].
+  void sortByOppositeValuesPerformance(
+      String positiveValue, String negativeValue) {
+    sort((a, b) {
+      var p1 = a.oppositeValuesPerformance(positiveValue, negativeValue);
+      var p2 = b.oppositeValuesPerformance(positiveValue, negativeValue);
+      return p1.compareTo(p2);
+    });
+  }
+
+  /// Sort sub lists by [ListAnswerExtension.oppositeValuesPerformanceUnnormalized].
+  void sortByOppositeValuesPerformanceUnnormalized(
+      String positiveValue, String negativeValue) {
+    sort((a, b) {
+      var p1 =
+          a.oppositeValuesPerformanceUnnormalized(positiveValue, negativeValue);
+      var p2 =
+          b.oppositeValuesPerformanceUnnormalized(positiveValue, negativeValue);
+      return p1.compareTo(p2);
+    });
+  }
+}
+
 /// Base class to analyse a [BayesianNetwork].
 abstract class Analyser {
   final BayesianNetwork network;
@@ -1327,7 +1481,7 @@ abstract class Analyser {
     var name = Value.resolveName(s, networkCache: network);
     var signal = Value.resolveSignal(name: s);
 
-    var node = network.getNode(name);
+    var node = network.getNodeByName(name);
 
     var valueName = node.getValueName(signal: signal);
 

lib/src/statistics_extension_num.dart

@@ -1,4 +1,5 @@
 import 'dart:math' as math;
+import 'dart:typed_data';
 
 import 'package:collection/collection.dart';
 import 'package:intl/intl.dart';
@@ -1043,17 +1044,21 @@ extension ListIntExtension on List<int> {
   }
 }
 
-extension ListComparablesExtension<T> on List<Comparable<T>> {
+extension IterableComparablesExtension<T> on Iterable<Comparable<T>> {
+  /// Compares two [Iterable]s.
+  ///
   /// Similar to [String.compareTo].
-  int compareWith(List<T> other) {
+  int compareWith(Iterable<T> other) {
+    if (identical(this, other)) return 0;
+
     final len1 = length;
     final len2 = other.length;
     final lim = math.min(len1, len2);
 
     var k = 0;
     while (k < lim) {
-      var c1 = this[k];
-      var c2 = other[k];
+      var c1 = elementAt(k);
+      var c2 = other.elementAt(k);
 
       var cmp = c1.compareTo(c2);
 
@@ -1065,3 +1070,30 @@ extension ListComparablesExtension<T> on List<Comparable<T>> {
     return len1 - len2;
   }
 }
+
+extension IterableUint8ListExtension on Iterable<Uint8List> {
+  /// Compares two [Iterable]s of [Uint8List].
+  ///
+  /// Similar to [String.compareTo].
+  int compareWith(Iterable<Uint8List> other) {
+    if (identical(this, other)) return 0;
+
+    int len1 = length;
+    int len2 = other.length;
+    int lim = math.min(len1, len2);
+
+    int k = 0;
+    while (k < lim) {
+      var c1 = elementAt(k);
+      var c2 = other.elementAt(k);
+
+      var cmp = c1.compareWith(c2);
+      if (cmp != 0) {
+        return cmp;
+      }
+
+      k++;
+    }
+    return len1 - len2;
+  }
+}

pubspec.yaml

@@ -1,6 +1,6 @@
 name: statistics
 description: Statistics package for easy and efficient data manipulation with built-in Bayesian Network (Bayes Net), many mathematical functions and tools.
-version: 1.0.15
+version: 1.0.16
 homepage: https://github.com/gmpassos/statistics
 
 environment:
@@ -9,10 +9,15 @@ environment:
 dependencies:
   intl: ^0.17.0
   collection: ^1.15.0
-  data_serializer: ^1.0.3
+  data_serializer: ^1.0.7
 
 dev_dependencies:
   lints: ^1.0.1
   test: ^1.17.9
   dependency_validator: ^3.1.0
   coverage: ^1.0.3
+
+#dependency_overrides:
+#  data_serializer:
+#    path: ../data_serializer
+