Merge pull request #119 from Behnke02/main

MERGE: Update to Logreg ML methods
akielaries · Dec 23, 2024 · 549446c · 549446c
2 parents e4f0b66 + 152137e
commit 549446c
Show file tree

Hide file tree

Showing 2 changed files with 64 additions and 1 deletion.
diff --git a/include/openGPMP/ml/logreg.hpp b/include/openGPMP/ml/logreg.hpp
@@ -88,7 +88,7 @@ class LogReg {
      * @return A vector of predicted probabilities
      */
     std::vector<double> predict(const std::vector<std::vector<double>> &X_test);
-
+    
     /**
      * @brief Computes the accuracy of the model on the given test data
      * @param X_test The feature matrix of the test data
@@ -98,6 +98,32 @@ class LogReg {
     double accuracy(const std::vector<std::vector<double>> &X_test,
                     const std::vector<int> &y_test);
 
+    /**
+     * @brief Computes the accuracy of the model on the given test data
+     * @param predictions The predicted labels of the test data
+     * @param y_test The true labels of the test data
+     * @return The accuracy of the model
+     */
+    double accuracy(const std::vector<double> &predictions,
+                    const std::vector<int> &y_test);
+
+    /*
+     * @brief Computes the precision of the model on the given test data
+     * @param X_test The feature matrix on the test data
+     * @param y_test The true labels of the test data
+     * @return The precision of the model
+     */
+    double precision(const std::vector<std::vector<double>> &X_test,
+                     const std::vector<int> &y_test);
+
+    /*
+     * @brief Computes the recall of the model on the given test data
+     * @param X_test The feature matrix on the test data
+     * @param y_test The true labels of the test data
+     * @return The recall of the model
+     */
+    double recall(const std::vector<std::vector<double>> &X_test,
+                     const std::vector<int> &y_test);
     /**
      * @brief Performs feature scaling on the input feature matrix
      * @param X The feature matrix to be scaled

diff --git a/modules/ml/logreg.cpp b/modules/ml/logreg.cpp
@@ -109,6 +109,43 @@ gpmp::ml::LogReg::accuracy(const std::vector<std::vector<double>> &X_test,
     return static_cast<double>(correct) / y_test.size();
 }
 
+double
+gpmp::ml::LogReg::accuracy(const std::vector<double> &predictions,
+                    const std::vector<int> &y_test){
+    int correct = 0;
+    for (size_t i = 0; i < predictions.size(); ++i) {
+        correct += ((predictions[i] >= 0.5 && y_test[i] == 1) ||
+                    (predictions[i] < 0.5 && y_test[i] == 0));
+    }
+    return static_cast<double>(correct) / y_test.size();
+}
+
+double
+gpmp::ml::LogReg::precision(const std::vector<std::vector<double>> &X_test,
+                             const std::vector<int> &y_test){
+    std::vector<double> predictions = predict(X_test);
+    int true_positives = 0;
+    int false_positives = 0;
+    for (size_t i = 0; i < predictions.size(); ++i) {
+        true_positives += (predictions[i] >= 0.5 && y_test[i] == 1);
+        false_positives += (predictions[i] >= 0.5 && y_test[i] == 0); 
+    }
+    return static_cast<double>(true_positives / (true_positives + false_positives));
+}
+
+double
+gpmp::ml::LogReg::recall(const std::vector<std::vector<double>> &X_test,
+                             const std::vector<int> &y_test){
+    std::vector<double> predictions = predict(X_test);
+    int true_positives = 0;
+    int false_negatives = 0;
+    for (size_t i = 0; i < predictions.size(); ++i) {
+        true_positives += (predictions[i] >= 0.5 && y_test[i] == 1);
+        false_negatives += (predictions[i] < 0.5 && y_test[i] == 1);
+    }
+    return static_cast<double>(true_positives / (true_positives + false_negatives));
+}
+
 double gpmp::ml::LogReg::sigmoid(double z) {
     return 1.0 / (1.0 + exp(-z));
 }