Changeset 1152

Timestamp:

Feb 26, 2008, 12:31:46 AM (16 years ago)

Author:

Peter

Message:

some docs for NBC, note docs that not yet match implementation refs #335

File:

• trunk/yat/classifier/NBC.h (modified) (3 diffs)

trunk/yat/classifier/NBC.h

@@ -40 +40 @@
 
   /**
-     @brief Naive Bayesian Classification.
+     @brief Naive Bayesian Classifier.
   
      Each class is modelled as a multinormal distribution with
@@ -52 +52 @@
   public:
     ///
-    /// Constructor taking the training data, the target vector, and
-    /// the distance measure as input.
+    /// Constructor taking the training data, the target vector.
     ///
     NBC(const MatrixLookup&, const Target&);
@@ -91 +90 @@
        estimated probability that sample belong to class \f$ j \f$
 
-       \f$ P_j = \frac{1}{Z}\prod_i{\frac{1}{\sigma_i}}
+       \f$ P_j = \frac{1}{Z}\prod_i{\frac{1}{\sqrt{2\pi\sigma_i^2}}}
        \exp(\frac{w_i(x_i-\mu_i)^2}{\sigma_i^2})\f$, where \f$ \mu_i
        \f$ and \f$ \sigma_i^2 \f$ are the estimated mean and variance,
-       respectively. If \a data is a MatrixLookup is equivalent to
-       using all weight equal to unity.
+       respectively. If a \f$ \sigma_i \f$ could not be estimated
+       during training, corresponding factor is set to unity, in other
+       words, that feature is ignored for the prediction of that
+       particular class. Z is chosen such that total probability, \f$
+       \sum P_j \f$, equals unity. If \a data is a MatrixLookup is
+       equivalent to using all weight equal to unity.
     */
     void predict(const DataLookup2D& data, utility::Matrix& res) const;