Activated 'float' math context

src/main/java/org/jpmml/tensorflow/DNNEstimator.java

@@ -26,6 +26,7 @@
 import com.google.common.primitives.Floats;
 import org.dmg.pmml.DataType;
 import org.dmg.pmml.Entity;
+import org.dmg.pmml.MathContext;
 import org.dmg.pmml.neural_network.NeuralInputs;
 import org.dmg.pmml.neural_network.NeuralLayer;
 import org.dmg.pmml.neural_network.NeuralNetwork;
@@ -51,7 +52,8 @@ protected NeuralNetwork encodeNeuralNetwork(TensorFlowEncoder encoder){
 		SavedModel savedModel = getSavedModel();
 
 		NeuralNetwork neuralNetwork = new NeuralNetwork()
-			.setActivationFunction(NeuralNetwork.ActivationFunction.RECTIFIER);
+			.setActivationFunction(NeuralNetwork.ActivationFunction.RECTIFIER)
+			.setMathContext(MathContext.FLOAT);
 
 		List<NodeDef> biasAdds = Lists.newArrayList(savedModel.getInputs(getHead(), "BiasAdd"));
 

src/main/java/org/jpmml/tensorflow/LinearClassifier.java

@@ -51,13 +51,13 @@ public RegressionModel encodeModel(TensorFlowEncoder encoder){
 		if(regressionTables.size() == 1){
 			categories = Arrays.asList("0", "1");
 
+			RegressionTable activeRegressionTable = regressionTables.get(0)
+				.setTargetCategory(categories.get(1));
+
 			RegressionTable passiveRegressionTable = new RegressionTable(0)
 				.setTargetCategory(categories.get(0));
 
 			regressionModel.addRegressionTables(passiveRegressionTable);
-
-			RegressionTable activeRegressionTable = regressionTables.get(0)
-				.setTargetCategory(categories.get(1));
 		} else
 
 		if(regressionTables.size() > 2){

src/main/java/org/jpmml/tensorflow/LinearEstimator.java

@@ -23,6 +23,7 @@
 import java.util.Map;
 
 import com.google.common.primitives.Floats;
+import org.dmg.pmml.MathContext;
 import org.dmg.pmml.regression.RegressionModel;
 import org.dmg.pmml.regression.RegressionTable;
 import org.jpmml.converter.CMatrixUtil;
@@ -143,7 +144,8 @@ public RegressionModel encodeRegressionModel(TensorFlowEncoder encoder){
 			}
 		}
 
-		RegressionModel regressionModel = new RegressionModel();
+		RegressionModel regressionModel = new RegressionModel()
+			.setMathContext(MathContext.FLOAT);
 
 		for(Equation equation : equations){
 			RegressionTable regressionTable = RegressionModelUtil.createRegressionTable(equation.getFeatures(), equation.getCoefficients(), equation.getIntercept());

src/test/java/org/jpmml/tensorflow/LinearRegressorTest.java

@@ -18,13 +18,12 @@
  */
 package org.jpmml.tensorflow;
 
-import org.jpmml.evaluator.PMMLEquivalence;
 import org.junit.Test;
 
 public class LinearRegressorTest extends EstimatorTest {
 
 	public LinearRegressorTest(){
-		super(new PMMLEquivalence(1e-6, 1e-6));
+		super(new TensorFlowEquivalence(2));
 	}
 
 	@Test

src/test/java/org/jpmml/tensorflow/TensorFlowEquivalence.java

@@ -0,0 +1,47 @@
+/*
+ * Copyright (c) 2017 Villu Ruusmann
+ *
+ * This file is part of JPMML-TensorFlow
+ *
+ * JPMML-TensorFlow is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * JPMML-TensorFlow is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with JPMML-TensorFlow.  If not, see <http://www.gnu.org/licenses/>.
+ */
+package org.jpmml.tensorflow;
+
+import org.dmg.pmml.DataType;
+import org.jpmml.evaluator.Computable;
+import org.jpmml.evaluator.EvaluatorUtil;
+import org.jpmml.evaluator.RealNumberEquivalence;
+import org.jpmml.evaluator.TypeUtil;
+
+public class TensorFlowEquivalence extends RealNumberEquivalence {
+
+	public TensorFlowEquivalence(int tolerance){
+		super(tolerance);
+	}
+
+	@Override
+	public boolean doEquivalent(Object expected, Object actual){
+
+		if(actual instanceof Computable){
+			actual = EvaluatorUtil.decode(actual);
+		} // End if
+
+		if(actual instanceof Number){
+			actual = TypeUtil.parseOrCast(DataType.FLOAT, actual);
+			expected = TypeUtil.parseOrCast(DataType.FLOAT, expected);
+		}
+
+		return super.doEquivalent(expected, actual);
+	}
+}