SVMClassification: Fix serialization

lib/src/LibSVM.cxx

@@ -339,7 +339,7 @@ Scalar LibSVM::runCrossValidation()
     totalError += (p_implementation_->problem_.y[i] - target[i]) * (p_implementation_->problem_.y[i] - target[i]) / size;
   }
 
-  LOGTRACE(OSS() << "LibSVM::runCrossValidation gamma=" << p_implementation_->parameter_.gamma << " C=" << p_implementation_->parameter_.C << " err=" << totalError);
+  LOGDEBUG(OSS() << "LibSVM::runCrossValidation gamma=" << p_implementation_->parameter_.gamma << " C=" << p_implementation_->parameter_.C << " err=" << totalError);
 
   return totalError;
 }

lib/src/SVMClassification.cxx

@@ -247,9 +247,9 @@ void SVMClassification::save(Advocate & adv) const
 void SVMClassification::load(Advocate & adv)
 {
   ClassifierImplementation::load(adv);
-  adv.saveAttribute( "tradeoffFactor_", tradeoffFactor_ );
-  adv.saveAttribute( "kernelParameter_", kernelParameter_ );
-  adv.saveAttribute( "accuracy_", accuracy_ );
+  adv.loadAttribute( "tradeoffFactor_", tradeoffFactor_ );
+  adv.loadAttribute( "kernelParameter_", kernelParameter_ );
+  adv.loadAttribute( "accuracy_", accuracy_ );
 }
 
 

python/doc/examples/plot_example2.py

@@ -31,9 +31,7 @@
 # we create dataIn and dataOut
 size = dataInOut.getSize()
 dataIn = dataInOut.getMarginal([1, 2])
-dataOut = ot.Indices(size)
-for i in range(size):
-    dataOut[i] = int(dataInOut[i, 0])
+dataOut = [int(dataInOut[i, 0]) for i in range(size)]
 
 # list of C parameter
 cp = [0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 1, 10, 100]

python/test/t_SVMClassification_multiclass.py

@@ -9,21 +9,9 @@
 path = os.path.abspath(os.path.dirname(__file__))
 dataInOut = ot.Sample.ImportFromCSVFile(os.path.join(path, "multiclass.csv"), ",")
 
-dataIn = ot.Sample(148, 4)
-dataOut = ot.Indices(148, 0)
-
-# we build the input Sample and the output Sample because we must separate
-# dataInOut
-for i in range(148):
-    a = dataInOut[i]
-    b = ot.Point(4)
-    b[0] = a[1]
-    b[1] = a[2]
-    b[2] = a[3]
-    b[3] = a[4]
-    dataIn[i] = b
-    dataOut[i] = int(a[0])
-
+size = len(dataInOut)
+dataIn = dataInOut.getMarginal([1, 2, 3, 4])
+dataOut = [int(dataInOut[i, 0]) for i in range(size)]
 
 # list of C parameter
 cp = [0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 1, 10, 100]

python/test/t_SVMClassification_std.py

@@ -7,20 +7,10 @@
 
 # we retrieve the sample from the file sample.csv
 path = os.path.abspath(os.path.dirname(__file__))
-dataInOut = ot.Sample().ImportFromCSVFile(os.path.join(path, "sample.csv"), ",")
-
-dataIn = ot.Sample(861, 2)
-dataOut = ot.Indices(861, 0)
-
-# we build the input Sample and the output Sample because we must separate
-# dataInOut
-for i in range(861):
-    a = dataInOut[i]
-    b = ot.Point(2)
-    b[0] = a[1]
-    b[1] = a[2]
-    dataIn[i] = b
-    dataOut[i] = int(a[0])
+dataInOut = ot.Sample.ImportFromCSVFile(os.path.join(path, "sample.csv"), ",")
+size = len(dataInOut)
+dataIn = dataInOut.getMarginal([1, 2])
+dataOut = [int(dataInOut[i, 0]) for i in range(size)]
 
 # list of C parameter
 cp = [0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 1, 10, 100]
@@ -40,3 +30,22 @@
 accuracy = algo.getAccuracy()
 print(accuracy)
 ott.assert_almost_equal(accuracy, 100.0)
+
+for i in range(size):
+    x = dataIn[i]
+    c = dataOut[i]
+    print(f"x={x} c={c} classify={algo.classify(x)} grade={algo.grade(x, c)} predict={algo.predict(x)}")
+
+if ot.PlatformInfo.HasFeature("libxml2"):
+    study = ot.Study()
+    fname = "study_classif.xml"
+    study.setStorageManager(ot.XMLStorageManager(fname))
+    study.add("algo", algo)
+    study.save()
+    study = ot.Study()
+    study.setStorageManager(ot.XMLStorageManager(fname))
+    study.load()
+    algo = otsvm.SVMClassification()
+    study.fillObject("algo", algo)
+    accuracy = algo.getAccuracy()
+    os.remove(fname)