Changeset 15023
 Timestamp:
 06/03/17 19:19:18 (7 weeks ago)
 File:

 1 edited
Legend:
 Unmodified
 Added
 Removed

trunk/sources/HeuristicLab.Algorithms.DataAnalysis.Glmnet/3.4/ElasticNetLinearRegression.cs
r14846 r15023 21 21 22 22 using System; 23 using System.Collections.Generic; 23 24 using System.Linq; 24 25 using System.Threading; … … 94 95 Results.Add(new Result("NMSE (test)", new DoubleValue(testNMSE))); 95 96 97 var ds = Problem.ProblemData.Dataset; 96 98 var allVariables = Problem.ProblemData.AllowedInputVariables.ToArray(); 97 98 var remainingVars = Enumerable.Range(0, allVariables.Length) 99 .Where(idx => !coeff[idx].IsAlmost(0.0)).Select(idx => allVariables[idx]) 100 .ToArray(); 101 var remainingCoeff = Enumerable.Range(0, allVariables.Length) 102 .Select(idx => coeff[idx]) 103 .Where(c => !c.IsAlmost(0.0)) 104 .ToArray(); 105 106 var tree = LinearModelToTreeConverter.CreateTree(remainingVars, remainingCoeff, coeff.Last()); 99 var doubleVariables = allVariables.Where(ds.VariableHasType<double>); 100 var factorVariableNames = allVariables.Where(ds.VariableHasType<string>); 101 var factorVariablesAndValues = ds.GetFactorVariableValues(factorVariableNames, Enumerable.Range(0, ds.Rows)); // must consider all factor values (in train and test set) 102 103 List<KeyValuePair<string, IEnumerable<string>>> remainingFactorVariablesAndValues = new List<KeyValuePair<string, IEnumerable<string>>>(); 104 List<double> factorCoeff = new List<double>(); 105 List<string> remainingDoubleVariables = new List<string>(); 106 List<double> doubleVarCoeff = new List<double>(); 107 108 { 109 int i = 0; 110 // find factor varibles & value combinations with nonzero coeff 111 foreach (var factorVarAndValues in factorVariablesAndValues) { 112 var l = new List<string>(); 113 foreach (var factorValue in factorVarAndValues.Value) { 114 if (!coeff[i].IsAlmost(0.0)) { 115 l.Add(factorValue); 116 factorCoeff.Add(coeff[i]); 117 } 118 i++; 119 } 120 if (l.Any()) remainingFactorVariablesAndValues.Add(new KeyValuePair<string, IEnumerable<string>>(factorVarAndValues.Key, l)); 121 } 122 // find double variables with nonzero coeff 123 foreach (var doubleVar in doubleVariables) { 124 if (!coeff[i].IsAlmost(0.0)) { 125 remainingDoubleVariables.Add(doubleVar); 126 doubleVarCoeff.Add(coeff[i]); 127 } 128 i++; 129 } 130 } 131 var tree = LinearModelToTreeConverter.CreateTree( 132 remainingFactorVariablesAndValues, factorCoeff.ToArray(), 133 remainingDoubleVariables.ToArray(), doubleVarCoeff.ToArray(), 134 coeff.Last()); 107 135 108 136 … … 140 168 var allowedVars = Problem.ProblemData.AllowedInputVariables.ToArray(); 141 169 var numNonZeroCoeffs = new int[nLambdas]; 142 for (int i = 0; i < nCoeff; i++) { 143 var coeffId = allowedVars[i]; 144 double sigma = Problem.ProblemData.Dataset.GetDoubleValues(coeffId).StandardDeviation(); 145 var path = Enumerable.Range(0, nLambdas).Select(r => Tuple.Create(lambda[r], coeff[r, i] * sigma)).ToArray(); 146 dataRows[i] = new IndexedDataRow<double>(coeffId, coeffId, path); 147 } 148 // add to coeffTable by total weight (larger area under the curve => more important); 149 foreach (var r in dataRows.OrderByDescending(r => r.Values.Select(t => t.Item2).Sum(x => Math.Abs(x)))) { 150 coeffTable.Rows.Add(r); 170 171 var ds = Problem.ProblemData.Dataset; 172 var doubleVariables = allowedVars.Where(ds.VariableHasType<double>); 173 var factorVariableNames = allowedVars.Where(ds.VariableHasType<string>); 174 var factorVariablesAndValues = ds.GetFactorVariableValues(factorVariableNames, Enumerable.Range(0, ds.Rows)); // must consider all factor values (in train and test set) 175 { 176 int i = 0; 177 foreach (var factorVariableAndValues in factorVariablesAndValues) { 178 foreach (var factorValue in factorVariableAndValues.Value) { 179 double sigma = ds.GetStringValues(factorVariableAndValues.Key) 180 .Select(s => s == factorValue ? 1.0 : 0.0) 181 .StandardDeviation(); // calc std dev of binary indicator 182 var path = Enumerable.Range(0, nLambdas).Select(r => Tuple.Create(lambda[r], coeff[r, i] * sigma)).ToArray(); 183 dataRows[i] = new IndexedDataRow<double>(factorVariableAndValues.Key + "=" + factorValue, factorVariableAndValues.Key + "=" + factorValue, path); 184 i++; 185 } 186 } 187 188 foreach (var doubleVariable in doubleVariables) { 189 double sigma = ds.GetDoubleValues(doubleVariable).StandardDeviation(); 190 var path = Enumerable.Range(0, nLambdas).Select(r => Tuple.Create(lambda[r], coeff[r, i] * sigma)).ToArray(); 191 dataRows[i] = new IndexedDataRow<double>(doubleVariable, doubleVariable, path); 192 i++; 193 } 194 // add to coeffTable by total weight (larger area under the curve => more important); 195 foreach (var r in dataRows.OrderByDescending(r => r.Values.Select(t => t.Item2).Sum(x => Math.Abs(x)))) { 196 coeffTable.Rows.Add(r); 197 } 151 198 } 152 199 … … 330 377 private static void PrepareData(IRegressionProblemData problemData, out double[,] trainX, out double[] trainY, 331 378 out double[,] testX, out double[] testY) { 332 333 379 var ds = problemData.Dataset; 334 trainX = ds.ToArray(problemData.AllowedInputVariables, problemData.TrainingIndices); 335 trainX = trainX.Transpose(); 336 trainY = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, 337 problemData.TrainingIndices) 338 .ToArray(); 339 testX = ds.ToArray(problemData.AllowedInputVariables, problemData.TestIndices); 340 testX = testX.Transpose(); 341 testY = problemData.Dataset.GetDoubleValues(problemData.TargetVariable, 342 problemData.TestIndices) 343 .ToArray(); 380 var targetVariable = problemData.TargetVariable; 381 var allowedInputs = problemData.AllowedInputVariables; 382 trainX = PrepareInputData(ds, allowedInputs, problemData.TrainingIndices); 383 trainY = ds.GetDoubleValues(targetVariable, problemData.TrainingIndices).ToArray(); 384 385 testX = PrepareInputData(ds, allowedInputs, problemData.TestIndices); 386 testY = ds.GetDoubleValues(targetVariable, problemData.TestIndices).ToArray(); 387 } 388 389 private static double[,] PrepareInputData(IDataset ds, IEnumerable<string> allowedInputs, IEnumerable<int> rows) { 390 var doubleVariables = allowedInputs.Where(ds.VariableHasType<double>); 391 var factorVariableNames = allowedInputs.Where(ds.VariableHasType<string>); 392 var factorVariables = ds.GetFactorVariableValues(factorVariableNames, Enumerable.Range(0, ds.Rows)); // must consider all factor values (in train and test set) 393 double[,] binaryMatrix = ds.ToArray(factorVariables, rows); 394 double[,] doubleVarMatrix = ds.ToArray(doubleVariables, rows); 395 var x = binaryMatrix.HorzCat(doubleVarMatrix); 396 return x.Transpose(); 344 397 } 345 398 }
Note: See TracChangeset
for help on using the changeset viewer.