Now portcullis filter accepts custom user-defined folders for the ini…

…tial training rules (see #41). Corrected deprecation warning in rule_filter.py

Also Resetting doc prefix

Corrected a bug that caused saltuarily segmentation faults when the size of the positive or the negative set were too small (smote.cc)

doc/Makefile

@@ -10,7 +10,7 @@ MKDIR_P = mkdir -p
 INSTALL = /usr/bin/install -c -m 644
 
 # Set autoconf variables
-prefix = /home/luca/workspace/portcullis/compiled
+prefix = /usr/local
 PACKAGE_TARNAME = portcullis
 top_srcdir = ..
 srcdir = .

lib/include/portcullis/ml/smote.hpp

@@ -44,6 +44,7 @@ class Smote {
  bool verbose;
 
  double* data;
+ size_t nelements;
  size_t rows;
  size_t cols;
 
@@ -52,7 +53,7 @@ class Smote {
 
 public:
 
- Smote(uint16_t defaultK, uint16_t _smoteness, uint16_t _threads, double* _data, size_t _rows, size_t _cols);
+ Smote(uint16_t defaultK, uint16_t _smoteness, uint16_t _threads, double* _data, size_t _nelements, size_t _rows, size_t _cols);
 
  ~Smote() {
  delete[] synthetic;
@@ -99,4 +100,4 @@ class Smote {
  void print(ostream& out) const;
 };
 }
-}
+}

lib/src/model_features.cc

@@ -273,7 +273,7 @@ portcullis::ml::ForestPtr portcullis::ml::ModelFeatures::trainInstance(const Jun
  }
  //cout << endl;
  }
- Smote smote(5, N, threads, nm, negData->getNumRows(), negData->getNumCols() - 1);
+ Smote smote(5, N, threads, nm, nelements, negData->getNumRows(), negData->getNumCols() - 1);
  smote.setVerbose(verbose);
  smote.execute();
  smote_rows = smote.getNbSynthRows();

lib/src/smote.cc

@@ -25,10 +25,11 @@ using portcullis::ml::KNN;
 
 #include <portcullis/ml/smote.hpp>
 
-portcullis::ml::Smote::Smote(uint16_t defaultK, uint16_t _smoteness, uint16_t _threads, double* _data, size_t _rows, size_t _cols) {
+portcullis::ml::Smote::Smote(uint16_t defaultK, uint16_t _smoteness, uint16_t _threads, double* _data, size_t _nelements, size_t _rows, size_t _cols) {
  data = _data;
  rows = _rows;
  cols = _cols;
+ nelements = _nelements;
  if (_rows < defaultK && _rows < 100)
  k = _rows;
  else
@@ -58,9 +59,16 @@ void portcullis::ml::Smote::execute() {
  const vector<uint32_t> nns = knn.getNNs(i);
  uint32_t nn = nns[igen(rng)]; // Nearest neighbour row index
  for (size_t j = 0; j < cols; j++) {
- double dif = data[(nn * cols) + j] - data[(i * cols) + j];
- double gap = dgen(rng);
- synthetic[(new_index * cols) + j] = data[(i * cols) + j] + gap * dif;
+ double dif;
+ double gap;
+ if ((nn * cols) + j <= nelements) {
+ dif = data[(nn * cols) + j] - data[(i * cols) + j]; // This is causing errors in some weird cases.
+ } else{
+ dif = 0;
+ }
+ gap = dgen(rng);
+ synthetic[(new_index * cols) + j] = data[(i * cols) + j] + gap * dif;
+
  }
  new_index++;
  N--;
@@ -84,4 +92,4 @@ void portcullis::ml::Smote::print(ostream& out) const {
  out << endl;
  }
  out << endl;
-}
+}