Randomly split data into training/testing set

bhklein
1 parent 1761523e
Showing 3 changed files with 51 additions and 26 deletions
openbr/openbr_plugin.cpp
openbr/openbr_plugin.h
openbr/plugins/core/crossvalidate.cpp
@@ -28,6 +28,7 @@
 #include <QtConcurrentRun>
 #include <algorithm>
 #include <iostream>
+#include <RandomLib/Random.hpp>
 #ifndef BR_EMBEDDED
 #include <QApplication>
@@ -528,28 +529,47 @@ QList&lt;int&gt; TemplateList::applyIndex(const QString &amp;propName, const QHash&lt;QString
     return result;
 }
-TemplateList TemplateList::partition(const QString &inputVariable) const
+TemplateList TemplateList::partition(const QString &inputVariable, int bootStrap) const
 {
-    const int crossValidate = Globals->crossValidate;
-    if (crossValidate < 2)
-        return *this;
-
     TemplateList partitioned = *this;
-    for (int i=partitioned.size()-1; i>=0; i--) {
-        // See CrossValidateTransform for description of these variables
-        if (partitioned[i].file.getBool("duplicatePartitions")) {
-            for (int j=crossValidate-1; j>=0; j--) {
-                Template duplicateTemplate = partitioned[i];
-                duplicateTemplate.file.set("Partition", j);
-                partitioned.insert(i+1, duplicateTemplate);
-            }
-        } else if (partitioned[i].file.getBool("allPartitions")) {
-            partitioned[i].file.set("Partition", -1);
-        } else {
-            if (!partitioned[i].file.contains(("Partition"))) {
-                const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
-                // Select the right 8 hex characters so that it can be represented as a 64 bit integer without overflow
-                partitioned[i].file.set("Partition", md5.toHex().right(8).toULongLong(0, 16) % crossValidate);
+    if (bootStrap > 0) {
+        // Use separate RNG from Common::randN() to avoid re-seeding the global RNG.
+        // rng is seeded with the inputVariable hash in order to maintain partition across br runs, given the same bootStrap seed.
+        RandomLib::Random rng;
+        for (int i=0; i < partitioned.size(); i++) {
+            const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
+            quint64 seed = md5.toHex().right(8).toULongLong(0, 16) + bootStrap;
+            rng.Reseed(seed);
+            float result = rng.FloatN();
+
+            // Roughly 2/3rd to training, 1/3rd to testing
+            if (result <= (2.0/3.0))
+                // Training
+                partitioned[i].file.set("Partition", QString::number(1));
+            else
+                partitioned[i].file.set("Partition", QString::number(0));
+        }
+    } else {
+        const int crossValidate = Globals->crossValidate;
+        if (crossValidate < 2)
+            return *this;
+
+        for (int i=partitioned.size()-1; i>=0; i--) {
+            // See CrossValidateTransform for description of these variables
+            if (partitioned[i].file.getBool("duplicatePartitions")) {
+                for (int j=crossValidate-1; j>=0; j--) {
+                    Template duplicateTemplate = partitioned[i];
+                    duplicateTemplate.file.set("Partition", j);
+                    partitioned.insert(i+1, duplicateTemplate);
+                }
+            } else if (partitioned[i].file.getBool("allPartitions")) {
+                partitioned[i].file.set("Partition", -1);
+            } else {
+                if (!partitioned[i].file.contains(("Partition"))) {
+                    const QByteArray md5 = QCryptographicHash::hash(partitioned[i].file.get<QString>(inputVariable).toLatin1(), QCryptographicHash::Md5);
+                    // Select the right 8 hex characters so that it can be represented as a 64 bit integer without overflow
+                    partitioned[i].file.set("Partition", md5.toHex().right(8).toULongLong(0, 16) % crossValidate);
+                }
             }
         }
     }
@@ -459,7 +459,7 @@ struct TemplateList : public QList&lt;Template&gt;
     BR_EXPORT static TemplateList relabel(const TemplateList &tl, const QString &propName, bool preserveIntegers);
     /*!< \brief Assign templates to folds partitions. */
-    BR_EXPORT TemplateList partition(const QString &inputVariable) const;
+    BR_EXPORT TemplateList partition(const QString &inputVariable, int bootStrap = 0) const;
     BR_EXPORT QList<int> indexProperty(const QString &propName, QHash<QString, int> * valueMap=NULL,QHash<int, QVariant> * reverseLookup = NULL) const;
     BR_EXPORT QList<int> indexProperty(const QString &propName, QHash<QString, int> &valueMap, QHash<int, QVariant> &reverseLookup) const;
@@ -56,8 +56,10 @@ class CrossValidateTransform : public MetaTransform
     Q_OBJECT
     Q_PROPERTY(QString description READ get_description WRITE set_description RESET reset_description STORED false)
     Q_PROPERTY(QString inputVariable READ get_inputVariable WRITE set_inputVariable RESET reset_inputVariable STORED false)
+    Q_PROPERTY(int bootStrap READ get_bootStrap WRITE set_bootStrap RESET reset_bootStrap STORED false)
     BR_PROPERTY(QString, description, "Identity")
     BR_PROPERTY(QString, inputVariable, "Label")
+    BR_PROPERTY(int, bootStrap, 0)
     // numPartitions copies of transform specified by description.
     QList<br::Transform*> transforms;
@@ -67,13 +69,12 @@ class CrossValidateTransform : public MetaTransform
     // is generally incorrect behavior.
     void train(const TemplateList &data)
     {
-        QList<int> partitions = data.partition(inputVariable).files().crossValidationPartitions();
-        const int numPartitions = Common::Max(partitions)+1;
-
+        QList<int> partitions = data.partition(inputVariable, bootStrap).files().crossValidationPartitions();
+        const int numPartitions = (bootStrap > 0) ? 1 : Common::Max(partitions)+1;
         while (transforms.size() < numPartitions)
             transforms.append(make(description));
-        if (numPartitions < 2) {
+        if (numPartitions < 2 && !(bootStrap > 0)) {
             transforms.first()->train(data);
             return;
         }
@@ -101,8 +102,12 @@ class CrossValidateTransform : public MetaTransform
     void project(const TemplateList &src, TemplateList &dst) const
     {
-        TemplateList partitioned = src.partition(inputVariable);
+        TemplateList partitioned = src.partition(inputVariable, bootStrap);
+        if (bootStrap > 0) {
+            transforms[0]->project(partitioned, dst);
+            return;
+        }
         for (int i=0; i<partitioned.size(); i++) {
             int partition = partitioned[i].file.get<int>("Partition", 0);
             transforms[partition]->project(partitioned, dst);