1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
4 | *
|
---|
5 | * This file is part of HeuristicLab.
|
---|
6 | *
|
---|
7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
8 | * it under the terms of the GNU General Public License as published by
|
---|
9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
10 | * (at your option) any later version.
|
---|
11 | *
|
---|
12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
15 | * GNU General Public License for more details.
|
---|
16 | *
|
---|
17 | * You should have received a copy of the GNU General Public License
|
---|
18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
19 | */
|
---|
20 | #endregion
|
---|
21 |
|
---|
22 | using System;
|
---|
23 | using System.Collections.Generic;
|
---|
24 | using System.Linq;
|
---|
25 | using HeuristicLab.Common;
|
---|
26 | using HeuristicLab.Core;
|
---|
27 | using HeuristicLab.Data;
|
---|
28 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
29 | using HeuristicLab.Optimization;
|
---|
30 | using HeuristicLab.Parameters;
|
---|
31 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
32 |
|
---|
33 | namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Regression {
|
---|
34 | [Item("SymbolicRegressionSingleObjectiveOSGAEvaluator", "An evaluator which tries to predict when a child will not be able to fullfil offspring selection criteria, to save evaluation time.")]
|
---|
35 | [StorableClass]
|
---|
36 | public class SymbolicRegressionSingleObjectiveOsgaEvaluator : SymbolicRegressionSingleObjectiveEvaluator {
|
---|
37 | private const string RelativeParentChildQualityThresholdParameterName = "RelativeParentChildQualityThreshold";
|
---|
38 | private const string RelativeFitnessEvaluationIntervalSizeParameterName = "RelativeFitnessEvaluationIntervalSize";
|
---|
39 | private const string ResultCollectionParameterName = "Results";
|
---|
40 | private const string AggregateStatisticsParameterName = "AggregateStatistics";
|
---|
41 | private const string ActualSelectionPressureParameterName = "SelectionPressure";
|
---|
42 | private const string UseAdaptiveQualityThresholdParameterName = "UseAdaptiveQualityThreshold";
|
---|
43 | private const string UseFixedEvaluationIntervalsParameterName = "UseFixedEvaluationIntervals";
|
---|
44 | private const string PreserveResultCompatibilityParameterName = "PreserveEvaluationResultCompatibility";
|
---|
45 |
|
---|
46 | #region parameters
|
---|
47 | public IFixedValueParameter<BoolValue> PreserveResultCompatibilityParameter {
|
---|
48 | get { return (IFixedValueParameter<BoolValue>)Parameters[PreserveResultCompatibilityParameterName]; }
|
---|
49 | }
|
---|
50 | public IFixedValueParameter<BoolValue> UseFixedEvaluationIntervalsParameter {
|
---|
51 | get { return (IFixedValueParameter<BoolValue>)Parameters[UseFixedEvaluationIntervalsParameterName]; }
|
---|
52 | }
|
---|
53 | public IFixedValueParameter<BoolValue> UseAdaptiveQualityThresholdParameter {
|
---|
54 | get { return (IFixedValueParameter<BoolValue>)Parameters[UseAdaptiveQualityThresholdParameterName]; }
|
---|
55 | }
|
---|
56 | public ILookupParameter<DoubleValue> ActualSelectionPressureParameter {
|
---|
57 | get { return (ILookupParameter<DoubleValue>)Parameters[ActualSelectionPressureParameterName]; }
|
---|
58 | }
|
---|
59 | public ILookupParameter<ResultCollection> ResultCollectionParameter {
|
---|
60 | get { return (ILookupParameter<ResultCollection>)Parameters[ResultCollectionParameterName]; }
|
---|
61 | }
|
---|
62 | public IValueParameter<BoolValue> AggregateStatisticsParameter {
|
---|
63 | get { return (IValueParameter<BoolValue>)Parameters[AggregateStatisticsParameterName]; }
|
---|
64 | }
|
---|
65 | public IValueParameter<IntMatrix> RejectedStatsParameter {
|
---|
66 | get { return (IValueParameter<IntMatrix>)Parameters["RejectedStats"]; }
|
---|
67 | }
|
---|
68 | public IValueParameter<IntMatrix> NotRejectedStatsParameter {
|
---|
69 | get { return (IValueParameter<IntMatrix>)Parameters["TotalStats"]; }
|
---|
70 | }
|
---|
71 | public IValueLookupParameter<DoubleValue> ComparisonFactorParameter {
|
---|
72 | get { return (ValueLookupParameter<DoubleValue>)Parameters["ComparisonFactor"]; }
|
---|
73 | }
|
---|
74 | public IFixedValueParameter<PercentValue> RelativeParentChildQualityThresholdParameter {
|
---|
75 | get { return (IFixedValueParameter<PercentValue>)Parameters[RelativeParentChildQualityThresholdParameterName]; }
|
---|
76 | }
|
---|
77 | public IFixedValueParameter<PercentValue> RelativeFitnessEvaluationIntervalSizeParameter {
|
---|
78 | get { return (IFixedValueParameter<PercentValue>)Parameters[RelativeFitnessEvaluationIntervalSizeParameterName]; }
|
---|
79 | }
|
---|
80 | public IScopeTreeLookupParameter<DoubleValue> ParentQualitiesParameter { get { return (IScopeTreeLookupParameter<DoubleValue>)Parameters["ParentQualities"]; } }
|
---|
81 | #endregion
|
---|
82 |
|
---|
83 | #region parameter properties
|
---|
84 | public bool AggregateStatistics {
|
---|
85 | get { return AggregateStatisticsParameter.Value.Value; }
|
---|
86 | set { AggregateStatisticsParameter.Value.Value = value; }
|
---|
87 | }
|
---|
88 | public bool PreserveResultCompatibility {
|
---|
89 | get { return PreserveResultCompatibilityParameter.Value.Value; }
|
---|
90 | set { PreserveResultCompatibilityParameter.Value.Value = value; }
|
---|
91 | }
|
---|
92 | public bool UseFixedEvaluationIntervals {
|
---|
93 | get { return UseFixedEvaluationIntervalsParameter.Value.Value; }
|
---|
94 | set { UseFixedEvaluationIntervalsParameter.Value.Value = value; }
|
---|
95 | }
|
---|
96 | public bool UseAdaptiveQualityThreshold {
|
---|
97 | get { return UseAdaptiveQualityThresholdParameter.Value.Value; }
|
---|
98 | set { UseAdaptiveQualityThresholdParameter.Value.Value = value; }
|
---|
99 | }
|
---|
100 | public double RelativeParentChildQualityThreshold {
|
---|
101 | get { return RelativeParentChildQualityThresholdParameter.Value.Value; }
|
---|
102 | set { RelativeParentChildQualityThresholdParameter.Value.Value = value; }
|
---|
103 | }
|
---|
104 | public double RelativeFitnessEvaluationIntervalSize {
|
---|
105 | get { return RelativeFitnessEvaluationIntervalSizeParameter.Value.Value; }
|
---|
106 | set { RelativeFitnessEvaluationIntervalSizeParameter.Value.Value = value; }
|
---|
107 | }
|
---|
108 | public IntMatrix RejectedStats {
|
---|
109 | get { return RejectedStatsParameter.Value; }
|
---|
110 | set { RejectedStatsParameter.Value = value; }
|
---|
111 | }
|
---|
112 | public IntMatrix TotalStats {
|
---|
113 | get { return NotRejectedStatsParameter.Value; }
|
---|
114 | set { NotRejectedStatsParameter.Value = value; }
|
---|
115 | }
|
---|
116 | #endregion
|
---|
117 |
|
---|
118 | public override bool Maximization {
|
---|
119 | get { return true; }
|
---|
120 | }
|
---|
121 |
|
---|
122 | public double AdjustedEvaluatedSolutions { get; set; }
|
---|
123 |
|
---|
124 | public SymbolicRegressionSingleObjectiveOsgaEvaluator() {
|
---|
125 | Parameters.Add(new ValueLookupParameter<DoubleValue>("ComparisonFactor", "Determines if the quality should be compared to the better parent (1.0), to the worse (0.0) or to any linearly interpolated value between them."));
|
---|
126 | Parameters.Add(new FixedValueParameter<PercentValue>(RelativeParentChildQualityThresholdParameterName, new PercentValue(0.9)));
|
---|
127 | Parameters.Add(new FixedValueParameter<PercentValue>(RelativeFitnessEvaluationIntervalSizeParameterName, new PercentValue(0.1)));
|
---|
128 | Parameters.Add(new LookupParameter<ResultCollection>(ResultCollectionParameterName));
|
---|
129 | Parameters.Add(new ScopeTreeLookupParameter<DoubleValue>("ParentQualities") { ActualName = "Quality" });
|
---|
130 | Parameters.Add(new ValueParameter<IntMatrix>("RejectedStats", new IntMatrix()));
|
---|
131 | Parameters.Add(new ValueParameter<IntMatrix>("TotalStats", new IntMatrix()));
|
---|
132 | Parameters.Add(new ValueParameter<BoolValue>(AggregateStatisticsParameterName, new BoolValue(false)));
|
---|
133 | Parameters.Add(new LookupParameter<DoubleValue>(ActualSelectionPressureParameterName));
|
---|
134 | Parameters.Add(new FixedValueParameter<BoolValue>(UseAdaptiveQualityThresholdParameterName, new BoolValue(false)));
|
---|
135 | Parameters.Add(new FixedValueParameter<BoolValue>(UseFixedEvaluationIntervalsParameterName, new BoolValue(false)));
|
---|
136 | Parameters.Add(new FixedValueParameter<BoolValue>(PreserveResultCompatibilityParameterName, new BoolValue(false)));
|
---|
137 | }
|
---|
138 |
|
---|
139 | [StorableHook(HookType.AfterDeserialization)]
|
---|
140 | private void AfterDeserialization() {
|
---|
141 | if (!Parameters.ContainsKey(ActualSelectionPressureParameterName))
|
---|
142 | Parameters.Add(new LookupParameter<DoubleValue>(ActualSelectionPressureParameterName));
|
---|
143 |
|
---|
144 | if (!Parameters.ContainsKey(UseAdaptiveQualityThresholdParameterName))
|
---|
145 | Parameters.Add(new FixedValueParameter<BoolValue>(UseAdaptiveQualityThresholdParameterName, new BoolValue(false)));
|
---|
146 |
|
---|
147 | if (!Parameters.ContainsKey(UseFixedEvaluationIntervalsParameterName))
|
---|
148 | Parameters.Add(new FixedValueParameter<BoolValue>(UseFixedEvaluationIntervalsParameterName, new BoolValue(false)));
|
---|
149 |
|
---|
150 | if (!Parameters.ContainsKey(PreserveResultCompatibilityParameterName))
|
---|
151 | Parameters.Add(new FixedValueParameter<BoolValue>(PreserveResultCompatibilityParameterName, new BoolValue(false)));
|
---|
152 | }
|
---|
153 |
|
---|
154 | [StorableConstructor]
|
---|
155 | protected SymbolicRegressionSingleObjectiveOsgaEvaluator(bool deserializing) : base(deserializing) { }
|
---|
156 |
|
---|
157 | protected SymbolicRegressionSingleObjectiveOsgaEvaluator(SymbolicRegressionSingleObjectiveOsgaEvaluator original, Cloner cloner) : base(original, cloner) { }
|
---|
158 |
|
---|
159 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
160 | return new SymbolicRegressionSingleObjectiveOsgaEvaluator(this, cloner);
|
---|
161 | }
|
---|
162 |
|
---|
163 | public override void ClearState() {
|
---|
164 | base.ClearState();
|
---|
165 | RejectedStats = new IntMatrix();
|
---|
166 | TotalStats = new IntMatrix();
|
---|
167 | AdjustedEvaluatedSolutions = 0;
|
---|
168 | }
|
---|
169 |
|
---|
170 | public override IOperation InstrumentedApply() {
|
---|
171 | var solution = SymbolicExpressionTreeParameter.ActualValue;
|
---|
172 | IEnumerable<int> rows = GenerateRowsToEvaluate();
|
---|
173 |
|
---|
174 | var interpreter = SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
|
---|
175 | var estimationLimits = EstimationLimitsParameter.ActualValue;
|
---|
176 | var problemData = ProblemDataParameter.ActualValue;
|
---|
177 | var applyLinearScaling = ApplyLinearScalingParameter.ActualValue.Value;
|
---|
178 |
|
---|
179 | double quality;
|
---|
180 | var parentQualities = ParentQualitiesParameter.ActualValue;
|
---|
181 |
|
---|
182 | // parent subscopes are not present during evaluation of the initial population
|
---|
183 | if (parentQualities.Length > 0) {
|
---|
184 | quality = Calculate(interpreter, solution, estimationLimits, problemData, rows);
|
---|
185 | } else {
|
---|
186 | quality = Calculate(interpreter, solution, estimationLimits.Lower, estimationLimits.Upper, problemData, rows, applyLinearScaling);
|
---|
187 | }
|
---|
188 | QualityParameter.ActualValue = new DoubleValue(quality);
|
---|
189 |
|
---|
190 | return base.InstrumentedApply();
|
---|
191 | }
|
---|
192 |
|
---|
193 | public static double Calculate(ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, ISymbolicExpressionTree solution, double lowerEstimationLimit, double upperEstimationLimit, IRegressionProblemData problemData, IEnumerable<int> rows, bool applyLinearScaling) {
|
---|
194 | IEnumerable<double> estimatedValues = interpreter.GetSymbolicExpressionTreeValues(solution, problemData.Dataset, rows);
|
---|
195 | IEnumerable<double> targetValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows);
|
---|
196 | OnlineCalculatorError errorState;
|
---|
197 |
|
---|
198 | double r;
|
---|
199 | if (applyLinearScaling) {
|
---|
200 | var rCalculator = new OnlinePearsonsRCalculator();
|
---|
201 | CalculateWithScaling(targetValues, estimatedValues, lowerEstimationLimit, upperEstimationLimit, rCalculator, problemData.Dataset.Rows);
|
---|
202 | errorState = rCalculator.ErrorState;
|
---|
203 | r = rCalculator.R;
|
---|
204 | } else {
|
---|
205 | IEnumerable<double> boundedEstimatedValues = estimatedValues.LimitToRange(lowerEstimationLimit, upperEstimationLimit);
|
---|
206 | r = OnlinePearsonsRCalculator.Calculate(targetValues, boundedEstimatedValues, out errorState);
|
---|
207 | }
|
---|
208 | if (errorState != OnlineCalculatorError.None) return double.NaN;
|
---|
209 | return r * r;
|
---|
210 | }
|
---|
211 |
|
---|
212 | private double Calculate(ISymbolicDataAnalysisExpressionTreeInterpreter interpreter, ISymbolicExpressionTree solution, DoubleLimit estimationLimits, IRegressionProblemData problemData, IEnumerable<int> rows) {
|
---|
213 | var estimatedValues = interpreter.GetSymbolicExpressionTreeValues(solution, problemData.Dataset, rows);
|
---|
214 | var targetValues = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, rows).ToList();
|
---|
215 | var parentQualities = ParentQualitiesParameter.ActualValue.Select(x => x.Value);
|
---|
216 | var minQuality = double.MaxValue;
|
---|
217 | var maxQuality = double.MinValue;
|
---|
218 |
|
---|
219 | foreach (var quality in parentQualities) {
|
---|
220 | if (minQuality > quality) minQuality = quality;
|
---|
221 | if (maxQuality < quality) maxQuality = quality;
|
---|
222 | }
|
---|
223 |
|
---|
224 | var comparisonFactor = ComparisonFactorParameter.ActualValue.Value;
|
---|
225 | var parentQuality = minQuality + (maxQuality - minQuality) * comparisonFactor;
|
---|
226 |
|
---|
227 | #region fixed intervals
|
---|
228 | if (UseFixedEvaluationIntervals) {
|
---|
229 | double threshold = parentQuality * RelativeParentChildQualityThreshold;
|
---|
230 | if (UseAdaptiveQualityThreshold) {
|
---|
231 | var actualSelectionPressure = ActualSelectionPressureParameter.ActualValue;
|
---|
232 | if (actualSelectionPressure != null)
|
---|
233 | threshold = parentQuality * (1 - actualSelectionPressure.Value / 100.0);
|
---|
234 | }
|
---|
235 | var estimatedEnumerator = estimatedValues.GetEnumerator();
|
---|
236 | var targetEnumerator = targetValues.GetEnumerator();
|
---|
237 |
|
---|
238 | var rcalc = new OnlinePearsonsRCalculator();
|
---|
239 | var trainingPartitionSize = problemData.TrainingPartition.Size;
|
---|
240 | var interval = (int)Math.Floor(trainingPartitionSize * RelativeFitnessEvaluationIntervalSize);
|
---|
241 |
|
---|
242 | var calculatedRows = 0;
|
---|
243 | #region aggregate statistics
|
---|
244 | if (AggregateStatistics) {
|
---|
245 | var trainingEnd = problemData.TrainingPartition.End;
|
---|
246 | var qualityPerInterval = new List<double>();
|
---|
247 | while (estimatedEnumerator.MoveNext() & targetEnumerator.MoveNext()) {
|
---|
248 | var estimated = estimatedEnumerator.Current;
|
---|
249 | var target = targetEnumerator.Current;
|
---|
250 | rcalc.Add(estimated, target);
|
---|
251 | ++calculatedRows;
|
---|
252 | if (calculatedRows % interval == 0 || calculatedRows == trainingPartitionSize) {
|
---|
253 | var r = rcalc.ErrorState == OnlineCalculatorError.None ? rcalc.R : 0d;
|
---|
254 | qualityPerInterval.Add(r * r);
|
---|
255 | }
|
---|
256 | }
|
---|
257 | double quality;
|
---|
258 | {
|
---|
259 | var r = rcalc.ErrorState != OnlineCalculatorError.None ? 0d : rcalc.R;
|
---|
260 | var actualQuality = r * r;
|
---|
261 | quality = actualQuality;
|
---|
262 | bool predictedRejected = false;
|
---|
263 |
|
---|
264 | calculatedRows = 0;
|
---|
265 | foreach (var q in qualityPerInterval) {
|
---|
266 | if (double.IsNaN(q) || !(q > threshold)) {
|
---|
267 | predictedRejected = true;
|
---|
268 | quality = q;
|
---|
269 | break;
|
---|
270 | }
|
---|
271 | ++calculatedRows;
|
---|
272 | }
|
---|
273 |
|
---|
274 | var actuallyRejected = !(actualQuality > parentQuality);
|
---|
275 |
|
---|
276 | if (RejectedStats.Rows == 0 || TotalStats.Rows == 0) {
|
---|
277 | RejectedStats = new IntMatrix(2, qualityPerInterval.Count);
|
---|
278 | RejectedStats.RowNames = new[] { "Predicted", "Actual" };
|
---|
279 | RejectedStats.ColumnNames = Enumerable.Range(1, RejectedStats.Columns).Select(x => string.Format("0-{0}", Math.Min(trainingEnd, x * interval)));
|
---|
280 | TotalStats = new IntMatrix(2, 2);
|
---|
281 | TotalStats.RowNames = new[] { "Predicted", "Actual" };
|
---|
282 | TotalStats.ColumnNames = new[] { "Rejected", "Not Rejected" };
|
---|
283 | }
|
---|
284 | // gather some statistics
|
---|
285 | if (predictedRejected) {
|
---|
286 | RejectedStats[0, calculatedRows]++;
|
---|
287 | TotalStats[0, 0]++;
|
---|
288 | } else {
|
---|
289 | TotalStats[0, 1]++;
|
---|
290 | }
|
---|
291 | if (actuallyRejected) {
|
---|
292 | TotalStats[1, 0]++;
|
---|
293 | } else {
|
---|
294 | TotalStats[1, 1]++;
|
---|
295 | }
|
---|
296 | if (predictedRejected && actuallyRejected) {
|
---|
297 | RejectedStats[1, calculatedRows]++;
|
---|
298 | }
|
---|
299 | }
|
---|
300 | return quality;
|
---|
301 | }
|
---|
302 | #endregion
|
---|
303 | else {
|
---|
304 | while (estimatedEnumerator.MoveNext() & targetEnumerator.MoveNext()) {
|
---|
305 | rcalc.Add(targetEnumerator.Current, estimatedEnumerator.Current);
|
---|
306 | ++calculatedRows;
|
---|
307 | if (calculatedRows % interval == 0 || calculatedRows == trainingPartitionSize) {
|
---|
308 | var q = rcalc.ErrorState != OnlineCalculatorError.None ? double.NaN : rcalc.R;
|
---|
309 | var quality = q * q;
|
---|
310 | if (!(quality > threshold)) {
|
---|
311 | AdjustedEvaluatedSolutions += (double)calculatedRows / problemData.TrainingPartition.Size;
|
---|
312 | return quality;
|
---|
313 | }
|
---|
314 | }
|
---|
315 | }
|
---|
316 | var r = rcalc.ErrorState != OnlineCalculatorError.None ? double.NaN : rcalc.R;
|
---|
317 | var actualQuality = r * r;
|
---|
318 | AdjustedEvaluatedSolutions += 1d;
|
---|
319 | return actualQuality;
|
---|
320 | }
|
---|
321 | #endregion
|
---|
322 | } else {
|
---|
323 | var lsc = new OnlineLinearScalingParameterCalculator();
|
---|
324 | var rcalc = new OnlinePearsonsRCalculator();
|
---|
325 | var interval = (int)Math.Round(RelativeFitnessEvaluationIntervalSize * problemData.TrainingPartition.Size);
|
---|
326 | var quality = 0d;
|
---|
327 | var calculatedRows = 0;
|
---|
328 |
|
---|
329 | var cache = PreserveResultCompatibility ? new List<double>(problemData.TrainingPartition.Size) : null;
|
---|
330 | foreach (var target in estimatedValues.Zip(targetValues, (e, t) => new { EstimatedValue = e, ActualValue = t })) {
|
---|
331 | if (cache != null)
|
---|
332 | cache.Add(target.EstimatedValue);
|
---|
333 |
|
---|
334 | lsc.Add(target.EstimatedValue, target.ActualValue);
|
---|
335 | rcalc.Add(target.EstimatedValue, target.ActualValue);
|
---|
336 |
|
---|
337 | calculatedRows++;
|
---|
338 |
|
---|
339 | if (calculatedRows % interval != 0) continue;
|
---|
340 |
|
---|
341 | var alpha = lsc.Alpha;
|
---|
342 | var beta = lsc.Beta;
|
---|
343 | if (lsc.ErrorState != OnlineCalculatorError.None) {
|
---|
344 | alpha = 0;
|
---|
345 | beta = 1;
|
---|
346 | }
|
---|
347 |
|
---|
348 | var calc = (OnlinePearsonsRCalculator)rcalc.Clone();
|
---|
349 | foreach (var t in targetValues.Skip(calculatedRows)) {
|
---|
350 | var s = (t - alpha) / beta; // scaled target
|
---|
351 | calc.Add(s, t); // add pair (scaled, target) to the calculator
|
---|
352 | }
|
---|
353 | var r = calc.ErrorState == OnlineCalculatorError.None ? calc.R : 0d;
|
---|
354 | quality = r * r;
|
---|
355 |
|
---|
356 | if (!(quality > parentQuality)) {
|
---|
357 | AdjustedEvaluatedSolutions += (double)calculatedRows / problemData.TrainingPartition.Size;
|
---|
358 | return quality;
|
---|
359 | }
|
---|
360 | }
|
---|
361 | if (PreserveResultCompatibility) {
|
---|
362 | // get quality for all the rows. to ensure reproducibility of results between this evaluator
|
---|
363 | // and the standard one, we calculate the quality in an identical way (otherwise the returned
|
---|
364 | // quality could be slightly off due to rounding errors (in the range 1e-15 to 1e-16)
|
---|
365 | var applyLinearScaling = ApplyLinearScalingParameter.ActualValue.Value;
|
---|
366 | double r;
|
---|
367 | OnlineCalculatorError calculatorError;
|
---|
368 |
|
---|
369 | if (applyLinearScaling) {
|
---|
370 | var alpha = lsc.Alpha;
|
---|
371 | var beta = lsc.Beta;
|
---|
372 | if (lsc.ErrorState != OnlineCalculatorError.None) {
|
---|
373 | alpha = 0;
|
---|
374 | beta = 1;
|
---|
375 | }
|
---|
376 | var boundedEstimatedValues = cache.Select(x => x * beta + alpha).LimitToRange(estimationLimits.Lower, estimationLimits.Upper);
|
---|
377 | r = OnlinePearsonsRCalculator.Calculate(boundedEstimatedValues, targetValues, out calculatorError);
|
---|
378 | } else {
|
---|
379 | var boundedEstimatedValues = cache.LimitToRange(estimationLimits.Lower, estimationLimits.Upper);
|
---|
380 | r = OnlinePearsonsRCalculator.Calculate(boundedEstimatedValues, targetValues, out calculatorError);
|
---|
381 | }
|
---|
382 | quality = calculatorError == OnlineCalculatorError.None ? r * r : 0d;
|
---|
383 | }
|
---|
384 | AdjustedEvaluatedSolutions += 1d;
|
---|
385 | return quality;
|
---|
386 | }
|
---|
387 | }
|
---|
388 |
|
---|
389 | public override double Evaluate(IExecutionContext context, ISymbolicExpressionTree tree, IRegressionProblemData problemData, IEnumerable<int> rows) {
|
---|
390 | SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = context;
|
---|
391 | EstimationLimitsParameter.ExecutionContext = context;
|
---|
392 | ApplyLinearScalingParameter.ExecutionContext = context;
|
---|
393 |
|
---|
394 | var interpreter = SymbolicDataAnalysisTreeInterpreterParameter.ActualValue;
|
---|
395 | var estimationLimits = EstimationLimitsParameter.ActualValue;
|
---|
396 | var applyLinearScaling = ApplyLinearScalingParameter.ActualValue.Value;
|
---|
397 |
|
---|
398 | double r2 = Calculate(interpreter, tree, estimationLimits.Lower, estimationLimits.Upper, problemData, rows, applyLinearScaling);
|
---|
399 |
|
---|
400 | SymbolicDataAnalysisTreeInterpreterParameter.ExecutionContext = null;
|
---|
401 | EstimationLimitsParameter.ExecutionContext = null;
|
---|
402 | ApplyLinearScalingParameter.ExecutionContext = null;
|
---|
403 |
|
---|
404 | return r2;
|
---|
405 | }
|
---|
406 | }
|
---|
407 | }
|
---|