Merge branch 'feature/function_separation' into develop

* feature/function_separation:
  bug fixes
  cmakelists update
  typo fix
  bug fix

GLP/Examples/CMakeLists.txt

@@ -11,8 +11,7 @@ IF(${CMAKE_SYSTEM_NAME} MATCHES "Linux")
   set_target_properties(gspan_pls_static PROPERTIES LINK_FLAGS "-static")
 ENDIF()
 
-install_targets(/bin gspan)
-install_targets(/bin gspan_static)
-install_targets(/bin gspan_pls_static)
+install(TARGETS gspan gspan_static gspan_pls_static
+        RUNTIME DESTINATION bin)
 
 add_subdirectory(gspls)

GLP/Examples/gspls/CMakeLists.txt

@@ -16,8 +16,6 @@ IF(${CMAKE_SYSTEM_NAME} MATCHES "Linux")
   set_target_properties(gspls-classify-static PROPERTIES LINK_FLAGS "-static")
 ENDIF()
 
-install_programs(/bin gspls)
-install_targets(/bin gspls-train)
-install_targets(/bin gspls-train-static)
-install_targets(/bin gspls-classify)
-install_targets(/bin gspls-classify-static)
+install(PROGRAMS gspls DESTINATION bin)
+install(TARGETS gspls-train gspls-train-static gspls-classify gspls-classify-static
+        RUNTIME DESTINATION bin)

GLP/Examples/gspls/gspls-train-main.cpp

@@ -34,7 +34,7 @@ string usage()
     "gspls-train is a part of GLP v1.0\n\n"
     "   Usage: gspls-train [options] [gsp file]\n\n"
     " Options: \n"
-    "          [--reg | --cla] regression or classification mode\n"
+    "          [--reg | --cla] regression or classification mode, default: regression\n"
     "          [-m] min frequency of common graphs, default: 2\n"
     "          [-L] max graph size for gspan mining, default: 10\n"
     "          [-n] max iterator number, default: 100\n"
@@ -77,12 +77,10 @@ int main(int argc, char* argv[])
         train->gspanResult = train->gspan(selectedCol);
 
         // spls
-        cout << "Train" << endl;
         MatrixXd features = get<MatrixXd>(train->gspanResult[SLGraphMiningResultTypeX]);
         train->splsResult = train->spls(features);
 
         // overfit detection
-        cout << "Validation" << endl;
         vector<Rule> rules = get< vector<Rule> >(train->gspanResult[SLGraphMiningResultTypeRules]);
 
         if (train->isOverfit(rules)){
@@ -97,7 +95,7 @@ int main(int argc, char* argv[])
     train->timeEnd();
 
     // write result to file
-    size_t best = i - param->resultHist.length;
+    size_t best = i == param->n ? i : i - param->resultHist.length;
     train->saveResults(best);
 
     return 0;

GLP/Examples/gspls/gspls-train.cpp

@@ -186,9 +186,9 @@ time_duration SLGsplsTrain::timeDuration()
     return time_duration(_timeEnd - _timeStart);
 }
 
-void SLGsplsTrain::setFileSuffix(string suffix)
+void SLGsplsTrain::setFilePrefix(string prefix)
 {
-    _fileSuffix = suffix;
+    _filePrefix = prefix;
 }
 
 SLGsplsTrain* SLGsplsTrain::initWithParam(TrainParameters& param)
@@ -217,11 +217,14 @@ SLGsplsTrain* SLGsplsTrain::initWithParam(TrainParameters& param)
     // load train gsp file
     if (param.respFile != NULL)
         EigenExt::loadMatrixFromFile(train->_trainRespMat, train->_param.respFile);
-
+    else
+        train->_trainRespMat = get<MatrixXd>(train->_gspls->getInnerValues(SLGraphMiningInnerValueY)[SLGraphMiningInnerValueY]);
+
     // calculate valid data length
     if (param.validLength < 0){
         param.validLength  = floor(train->_trainRespMat.rows() * VALID_RATIO);
     }
+
     // setup response data
     train->_validRespMat = train->_trainRespMat.bottomRows(param.validLength);
     train->_trainRespMat = train->_trainRespMat.topRows(train->_trainRespMat.rows() - param.validLength);
@@ -246,15 +249,15 @@ SLGsplsTrain* SLGsplsTrain::initWithParam(TrainParameters& param)
     // graph changed, rebuild dfs tree;
     trainGspan.rebuildDFSTree();
 
-    // outout file suffix
-    string suffix = (format("gspls_m%d_L%d_n%d_k%d_t%d_") %
+    // outout file prefix
+    string prefix = (format("gspls_m%d_L%d_n%d_k%d_t%d_") %
                             param.minsup                  %
                             param.maxpat                  %
                             param.n                       %
                             param.topk                    %
                             param.resultHist.length
                      ).str();
-    train->setFileSuffix(suffix);
+    train->setFilePrefix(prefix);
     return train;
 }
 
@@ -267,6 +270,8 @@ SLGraphMiningResult SLGsplsTrain::gspan(MatrixXd& selectedColumn)
 
 SLModelResult SLGsplsTrain::spls(MatrixXd& feature)
 {
+    cout << "Train" << endl;
+
     SLModelResult result = _gspls->train(feature, _trainRespMat, _param.mode->getResultType());
 
     long rows    = feature.rows();
@@ -288,6 +293,11 @@ bool SLGsplsTrain::isOverfit(vector<Rule> rules)
 {
     _param.resultHist.push_front(make_pair(gspanResult, splsResult));
 
+    // do not detect overfit if validation data was not been set
+    if (_param.validLength == 0) return false;
+
+    cout << "Validation" << endl;
+
     MatrixXd result(_validTransaction.size(), rules.size());
     _validGspan->buildDarts(rules);
     for (size_t i = 0; i < _validTransaction.size(); ++i) {
@@ -321,10 +331,10 @@ void SLGsplsTrain::saveResults(size_t index)
         resultPair = _param.resultHist.back();
 
     // best index
-    cout << "Best: n = " << index << endl;
+    cout << "Best: n = " << best << endl;
 
     // beta
-    ofstream outBeta((_fileSuffix+"Beta.txt").c_str(), ios::out);
+    ofstream outBeta((_filePrefix+"Beta.txt").c_str(), ios::out);
     outBeta.precision(12);
     outBeta.flags(ios::left);
     outBeta << resultPair.second[SLModelResultTypeBeta] << endl;
@@ -333,7 +343,7 @@ void SLGsplsTrain::saveResults(size_t index)
     // dfs
     cout << "DFS" << endl;
     vector<Rule> DFSes = get< vector<Rule> >(resultPair.first[SLGraphMiningResultTypeRules]);
-    ofstream outDFS((_fileSuffix+"DFS.txt").c_str(), ios::out);
+    ofstream outDFS((_filePrefix+"DFS.txt").c_str(), ios::out);
     for ( size_t i = 0; i < DFSes.size(); ++i )
     {
         cout   << DFSes[i].dfs << endl;

GLP/Examples/gspls/gspls-train.hpp

@@ -131,7 +131,7 @@ class SLGsplsTrain
     bool _isOverfit;
 
     ptime _timeStart, _timeEnd;
-    string _fileSuffix;
+    string _filePrefix;
 
     SLGlpProduct<SLSparsePls, SLGspan>* _gspls;
 
@@ -154,7 +154,7 @@ class SLGsplsTrain
     void timeEnd();
     time_duration timeDuration();
 
-    void setFileSuffix(string);
+    void setFilePrefix(string);
 
     MatrixXd& getTrainMat();
     MatrixXd& getTrainRespMat();

GLP/Library/CMakeLists.txt

@@ -13,7 +13,9 @@ add_library(glp_static STATIC ${SOURCES})
 set_target_properties(glp_static PROPERTIES OUTPUT_NAME "glp")
 set_target_properties(glp_static PROPERTIES CLEAN_DIRECT_OUTPUT 1)
 
-install_targets(/lib glp glp_static)
+install(TARGETS glp glp_static
+        LIBRARY DESTINATION lib
+        ARCHIVE DESTINATION lib/static)
 
 install(DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/
         DESTINATION include

README.md

@@ -10,7 +10,7 @@ Eigen 3.1.2 or higher.
 Boost 1.4.2 or higher.
 
 # Installation
-make
+make  
 make install
 
 # License