diff --git a/src/methods/naive_bayes_classifier/naive_bayes_classifier.cpp b/src/methods/naive_bayes_classifier/naive_bayes_classifier.cpp
index e06b44f0cb167808bf5b6071dbee7961ed4f4409..5502c2dc10676338f80c5842571ca50272451382 100644
--- a/src/methods/naive_bayes_classifier/naive_bayes_classifier.cpp
+++ b/src/methods/naive_bayes_classifier/naive_bayes_classifier.cpp
@@ -17,6 +17,8 @@ using namespace mlpack::naive_bayes;
// Global Variable of the NaiveBayesClassifier object so it can be accessed from all functions
NaiveBayesClassifier<mat> naiveBayesClassifier;
+bool isModelTrained = false;
+int trainedDimensionality = 0;
// input: const MatType & data,
@@ -50,11 +52,18 @@ void initModelWithTrain(float *dataMatArr, SP_integer dataMatSize, SP_integer da
{
naiveBayesClassifier = NaiveBayesClassifier(data, labelsVector, numClasses, (incrementalVariance == 1), epsilon);
}
+ catch(const std::out_of_range& e)
+ {
+ raisePrologSystemExeption("The given Labels dont fit the format [0,Numclasses-1]!");
+ return;
+ }
catch(const std::exception& e)
{
raisePrologSystemExeption(e.what());
return;
}
+ isModelTrained = true;
+ trainedDimensionality = data.n_rows;
}
@@ -76,6 +85,8 @@ void initModelNoTrain(SP_integer dimensionality, SP_integer numClasses, double e
raisePrologSystemExeption(e.what());
return;
}
+ isModelTrained = false;
+ trainedDimensionality = dimensionality;
}
/*
@@ -119,8 +130,20 @@ void classifyMatrix(float *dataMatArr, SP_integer dataMatSize, SP_integer dataMa
float **predicArr, SP_integer *predicArrSize,
float **probsMatArr, SP_integer *probsMatColNum, SP_integer *probsMatRowNum)
{
+ if (!isModelTrained)
+ {
+ raisePrologSystemExeption("The Model is not trained!");
+ return;
+ }
// convert the Prolog arrays to arma::mat
mat data = convertArrayToMat(dataMatArr, dataMatSize, dataMatRowNum);
+ // check if data Dim fits trained Dim
+ if (data.n_rows != trainedDimensionality)
+ {
+ raisePrologSystemExeption("The Datapoints Dimensionality doesnt fit the trained Dimensionality!");
+ return;
+ }
+
// get the ReturnVector
Row<size_t> predicReturnVector;
@@ -153,6 +176,11 @@ void classifyMatrix(float *dataMatArr, SP_integer dataMatSize, SP_integer dataMa
//
void means(float **meansMatArr, SP_integer *meansMatColNum, SP_integer *meansMatRowNum)
{
+ if (!isModelTrained)
+ {
+ raisePrologSystemExeption("The Model is not trained!");
+ return;
+ }
// get the ReturnMat
mat meansReturnMat = naiveBayesClassifier.Means();
@@ -169,6 +197,11 @@ void means(float **meansMatArr, SP_integer *meansMatColNum, SP_integer *meansMat
//
void probabilities(float **probsMatArr, SP_integer *probsMatColNum, SP_integer *probsMatRowNum)
{
+ if (!isModelTrained)
+ {
+ raisePrologSystemExeption("The Model is not trained!");
+ return;
+ }
// get the ReturnMat
mat probsReturnMat = naiveBayesClassifier.Probabilities();
@@ -199,6 +232,12 @@ void trainMatrix(float *dataMatArr, SP_integer dataMatSize, SP_integer dataMatRo
raisePrologSystemExeption("The number of data points does not match the number of labels!");
return;
}
+ // check if data Dim fits trained Dim
+ if (data.n_rows != trainedDimensionality)
+ {
+ raisePrologSystemExeption("The Datapoints Dimensionality doesnt fit the trained Dimensionality!");
+ return;
+ }
// convert the Prolog array to arma::rowvec
Row<size_t> labelsVector = convertArrayToVec(labelsArr, labelsArrSize);
@@ -218,6 +257,8 @@ void trainMatrix(float *dataMatArr, SP_integer dataMatSize, SP_integer dataMatRo
raisePrologSystemExeption(e.what());
return;
}
+
+ isModelTrained = true;
}
@@ -231,7 +272,13 @@ void trainPoint(float *pointArr, SP_integer pointArrSize,
SP_integer label)
{
// convert the Prolog array to arma::rowvec
- rowvec pointVector = convertArrayToRowvec(pointArr, pointArrSize);
+ vec pointVector = convertArrayToColvec(pointArr, pointArrSize);
+ // check if data Dim fits trained Dim
+ if (pointVector.size() != trainedDimensionality)
+ {
+ raisePrologSystemExeption("The Datapoints Dimensionality doesnt fit the trained Dimensionality!");
+ return;
+ }
if(label < 0)
{
@@ -249,6 +296,7 @@ void trainPoint(float *pointArr, SP_integer pointArrSize,
raisePrologSystemExeption(e.what());
return;
}
+ isModelTrained = true;
}
@@ -259,6 +307,12 @@ void trainPoint(float *pointArr, SP_integer pointArrSize,
//
void variances(float **variancesMatArr, SP_integer *variancesMatColNum, SP_integer *variancesMatRowNum)
{
+ if (!isModelTrained)
+ {
+ raisePrologSystemExeption("The Model is not trained!");
+ return;
+ }
+
// get the ReturnMat
mat variancesReturnMat = naiveBayesClassifier.Variances();
diff --git a/src/methods/naive_bayes_classifier/naive_bayes_classifier.pl b/src/methods/naive_bayes_classifier/naive_bayes_classifier.pl
index ea840a47c57cbd13636622eae2544ea90b96b094..0d799fb7da89822e141e2d7b182b21ba2508a282 100644
--- a/src/methods/naive_bayes_classifier/naive_bayes_classifier.pl
+++ b/src/methods/naive_bayes_classifier/naive_bayes_classifier.pl
@@ -38,6 +38,8 @@
%% Initializes the classifier as per the input and then trains it by calculating the sample mean and variances.
%%
nbc_initModelWithTrain(DataList, DataRows, LabelsList, NumClasses, IncrementalVar, Epsilon) :-
+ NumClasses >= 0,
+ Epsilon >= 0,
convert_list_to_float_array(DataList, DataRows, array(Xsize, Xrownum, X)),
convert_list_to_float_array(LabelsList, array(Ysize, Y)),
initModelWithTrainI(X, Xsize, Xrownum, Y, Ysize, NumClasses, IncrementalVar, Epsilon).
@@ -50,8 +52,8 @@ foreign(initModelWithTrain, c, initModelWithTrainI( +pointer(float_array), +
%% --Input--
+%% int dimensionality => 1,
%% int numClasses => 0,
-%% bool incrementalVariance => (1)true / (0)false => false,
%% float32 epsilon => 1e-10
%%
%% --Output--
@@ -60,10 +62,11 @@ foreign(initModelWithTrain, c, initModelWithTrainI( +pointer(float_array), +
%% Initialize the Naive Bayes classifier without performing training.
%% All of the parameters of the model will be initialized to zero. Be sure to use train before calling classify, otherwise the results may be meaningless.
%%
-nbc_initModelNoTrain(NumClasses, IncrementalVar, Epsilon) :-
+nbc_initModelNoTrain(Dimensionality, NumClasses, Epsilon) :-
+ Dimensionality > 0,
NumClasses >= 0,
Epsilon >= 0,
- initModelNoTrainI(NumClasses, IncrementalVar, Epsilon).
+ initModelNoTrainI(Dimensionality, NumClasses, Epsilon).
foreign(initModelNoTrain, c, initModelNoTrainI( +integer,
+integer,
diff --git a/src/methods/naive_bayes_classifier/naive_bayes_classifier_test.pl b/src/methods/naive_bayes_classifier/naive_bayes_classifier_test.pl
index 541088cb90c25ba3c1889fb35f48380ffb99ebbb..7db8158f4d11593cf93f8bc3f870c48f665fd91b 100644
--- a/src/methods/naive_bayes_classifier/naive_bayes_classifier_test.pl
+++ b/src/methods/naive_bayes_classifier/naive_bayes_classifier_test.pl
@@ -7,7 +7,7 @@
:- use_module('../../helper_files/helper.pl').
reset_Model_NoTrain :-
- nbc_initModelNoTrain(2, 0, 0.000001).
+ nbc_initModelNoTrain(3, 0, 0.000001).
reset_Model_WithTrain :-
nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1,0,1], 2, 0, 0.000001).
@@ -18,18 +18,21 @@ reset_Model_WithTrain :-
:- begin_tests(nbc_initModelNoTrain).
%% Failure Tests
+
+test(nbc_InitModelNoTrain_Negative_Dimensionality, fail) :-
+ nbc_initModelNoTrain(-3, 0, 0.000001).
test(nbc_InitModelNoTrain_Negative_NumClasses, fail) :-
- nbc_initModelNoTrain(-2, 0, 0.000001).
+ nbc_initModelNoTrain(3, -1, 0.000001).
test(nbc_InitModelNoTrain_Negative_Epsilon, fail) :-
- nbc_initModelNoTrain(2, 0, -0.0001).
+ nbc_initModelNoTrain(3, 0, -0.0001).
%% Successful Tests
test(nbc_InitModelNoTrain_Normal_Use) :-
- nbc_initModelNoTrain(2, 0, 0.000001).
+ nbc_initModelNoTrain(3, 0, 0.000001).
:- end_tests(nbc_initModelNoTrain).
@@ -48,13 +51,14 @@ test(nbc_InitModelWithTrain_Negative_NumClasses, fail) :-
test(nbc_InitModelWithTrain_Negative_Epsilon, fail) :-
nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1,0,1], 2, 0, -0.000001).
-test(nbc_InitModelWithTrain_Too_Short_Label, [error(_,system_error('Error'))]) :-
+
+test(nbc_InitModelWithTrain_Too_Short_Label, [error(_,system_error('The number of data points does not match the number of labels!'))]) :-
nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1], 2, 0, 0.000001).
-test(nbc_InitModelWithTrain_Too_Long_Label, [error(_,system_error('Error'))]) :-
+test(nbc_InitModelWithTrain_Too_Long_Label, [error(_,system_error('The number of data points does not match the number of labels!'))]) :-
nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1,0,1,0,1], 2, 0, 0.000001).
-test(nbc_InitModelWithTrain_Too_Many_Label_Classes, [error(_,system_error('Error'))]) :-
+test(nbc_InitModelWithTrain_Too_Many_Label_Classes, [error(_,system_error('The given Labels dont fit the format [0,Numclasses-1]!'))]) :-
nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1,2,3], 2, 0, 0.000001).
@@ -79,11 +83,11 @@ test(nbc_InitModelWithTrain_CSV_Input) :-
%% Failure Tests
-test(nbc_ClassifyMatrix_Before_Train, [error(_,system_error('Error'))]) :-
+test(nbc_ClassifyMatrix_Before_Train, [error(_,system_error('The Model is not trained!'))]) :-
reset_Model_NoTrain,
nbc_classifyMatrix([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, _, _, _).
-test(nbc_ClassifyMatrix_Diffrent_Dims_To_Train, [error(_,system_error('Error'))]) :-
+test(nbc_ClassifyMatrix_Diffrent_Dims_To_Train, [error(_,system_error('The Datapoints Dimensionality doesnt fit the trained Dimensionality!'))]) :-
reset_Model_WithTrain,
nbc_classifyMatrix([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 4, _, _, _).
@@ -99,7 +103,7 @@ test(nbc_ClassifyMatrix_Normal_Use) :-
print(ProbabilitiesList).
test(nbc_ClassifyMatrix_CSV_Input) :-
- reset_Model_WithTrain,
+ nbc_initModelWithTrain([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 4, [0,1,0], 2, 0, 0.000001),
open('src/data_csv/iris2.csv', read, File),
take_csv_row(File, skipFirstRow,10, Data),
nbc_classifyMatrix(Data, 4, PredictionList, _, ProbabilitiesList),
@@ -118,7 +122,7 @@ test(nbc_ClassifyMatrix_CSV_Input) :-
:- begin_tests(nbc_means).
%% Failure Tests
-test(nbc_Means_Before_Train, [error(_,system_error('Error'))]) :-
+test(nbc_Means_Before_Train, [error(_,system_error('The Model is not trained!'))]) :-
reset_Model_NoTrain,
nbc_means(MeansList, _),
print('\nMeans: '),
@@ -143,7 +147,7 @@ test(nbc_Means_Normal_Use) :-
:- begin_tests(nbc_probabilities).
%% Failure Tests
-test(nbc_Probabilities_Before_Train, [error(_,system_error('Error'))]) :-
+test(nbc_Probabilities_Before_Train, [error(_,system_error('The Model is not trained!'))]) :-
reset_Model_NoTrain,
nbc_probabilities(ProbabilitiesList, _),
print('\nProbabilities: '),
@@ -185,7 +189,7 @@ test(nbc_TrainMatrix_Too_Many_Label_Classes, [error(_,system_error('The given La
reset_Model_NoTrain,
nbc_trainMatrix([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 3, [0,1,2,3], 2, 0).
-test(nbc_TrainMatrix_After_InitTrain, [error(_,system_error('addition: incompatible matrix dimensions: 3x1 and 4x1'))]) :-
+test(nbc_TrainMatrix_After_InitTrain, [error(_,system_error('The Datapoints Dimensionality doesnt fit the trained Dimensionality!'))]) :-
reset_Model_WithTrain,
nbc_trainMatrix([5.1,3.5,1.4,4.9,3.0,1.4,4.7,3.2,1.3,4.6,3.1,1.5], 4, [0,1,0], 2, 0).
@@ -198,7 +202,7 @@ test(nbc_TrainMatrix_Normal_Use) :-
test(nbc_TrainMatrix_CSV_Input) :-
- reset_Model_NoTrain,
+ nbc_initModelNoTrain(4, 0, 0.000001),
open('src/data_csv/iris2.csv', read, File),
take_csv_row(File, skipFirstRow,10, Data),
nbc_trainMatrix(Data, 4, [0,1,0,1,1,0,1,1,1,0], 2, 0).
@@ -214,11 +218,15 @@ test(nbc_TrainMatrix_CSV_Input) :-
%% Failure Tests
-test(nbc_TrainPoint_Bad_Label, [error(_,system_error('Error'))]) :-
+test(nbc_TrainPoint_Bad_Label, [error(_,system_error('The given Label should be positive!'))]) :-
reset_Model_NoTrain,
nbc_trainPoint([5.1,3.5,1.4], -5).
-test(nbc_TrainPoint_Too_Long_Point, [error(_,system_error('Error'))]) :-
+test(nbc_TrainPoint_Different_Dim, [error(_,system_error('The Datapoints Dimensionality doesnt fit the trained Dimensionality!'))]) :-
+ reset_Model_NoTrain,
+ nbc_trainPoint([5.1,3.5,1.4,3.5], 0).
+
+test(nbc_TrainPoint_Too_Long_Point, [error(_,system_error('The Datapoints Dimensionality doesnt fit the trained Dimensionality!'))]) :-
reset_Model_WithTrain,
nbc_trainPoint([5.1,3.5,1.4,3.5], 0).
@@ -243,7 +251,7 @@ test(nbc_TrainPoint_After_InitTrain) :-
:- begin_tests(nbc_variances).
%% Failure Tests
-test(nbc_Variances_Before_Train, [error(_,system_error('Error'))]) :-
+test(nbc_Variances_Before_Train, [error(_,system_error('The Model is not trained!'))]) :-
reset_Model_NoTrain,
nbc_variances(VariancesList, _),
print('\nVariances: '),