Naive Bayes Classifier

The Naive Bayes Classifier is a probabilistic classifier, which uses Naive Bayes theorem. The Naive Bayes Classifier tries by using probability to determine which label an object you try to classify belongs to. This classifier has been trained with documents originating from characters wiki pages, where the species for that character is the label. So the feature set for each character is contained the words taken from their Wiki page and labeled with their species.

The classifier claims the accuracy is 65.6666666667 %, and you can try it yourself below, where you can try classify a single word or a sentence.

Classification result

 : 

The data table below shows the 10 most informative features in the classifier.


                  menace = True           Otolla : Human  =    110.0 : 1.0
                resulted = True           Sith ( : Human  =    110.0 : 1.0
                     led = True           Lannik : Human  =     73.3 : 1.0
                  toward = True           Geonos : Human  =     73.3 : 1.0
               palpatine = True            Duros : Human  =     26.4 : 1.0
               skywalker = True            Duros : Human  =     26.4 : 1.0
                    file = True           Yuuzha : Human  =     20.0 : 1.0
                  recent = True           unknow : Human  =     12.2 : 1.0
                     han = True           Yuuzha : Human  =     12.0 : 1.0
                   would = True           unknow : Human  =      7.3 : 1.0

The table below shows the test set and the results of testing the classifier with it. The test set does not contain words, which the classifier was trained with and the result of the test gave 67 % Correct, 32 % Incorrect guesses.