7:00 – 3:30 ASRC

• Adding in Wayne’s comments.
• Got the Corpus generating arff files for BagOfWords and TF-IDF.
• Here’s the result for NaiveBayes on the first four chapters of Mobey Dick

• Correctly Classified Instances 3 75 %
  Incorrectly Classified Instances 1 25 %
  Kappa statistic 0.6667
  Mean absolute error 0.125 
  Root mean squared error 0.3536
  Relative absolute error 29.1667 %
  Root relative squared error 71.4435 %
  Total Number of Instances 4 
  
  === Detailed Accuracy By Class ===
  
   TP Rate FP Rate Precision Recall F-Measure MCC ROC Area PRC Area Class
   1.000 0.000 1.000 1.000 1.000 1.000 1.000 1.000 1_-_Loomings
   0.000 0.000 0.000 0.000 0.000 0.000 0.500 0.250 3_-_The_Spouter_Inn
   1.000 0.333 0.500 1.000 0.667 0.577 0.833 0.500 2_-_The_Carpet_Bag
   1.000 0.000 1.000 1.000 1.000 1.000 1.000 1.000 4_-_The_Counterpane
  Weighted Avg. 0.750 0.083 0.625 0.750 0.667 0.644 0.833 0.688 
  
  === Confusion Matrix ===
  
   a b c d <-- classified as
   1 0 0 0 | a = 1_-_Loomings
   0 0 1 0 | b = 3_-_The_Spouter_Inn
   0 0 1 0 | c = 2_-_The_Carpet_Bag
   0 0 0 1 | d = 4_-_The_Counterpane

• This worked really well: weka.classifiers.functions.Logistic -R 1.0E-8 -M -1 -num-decimal-places 4
• And comparing Jack London stories to Edgar Allen Poe stories works with a corpus of six stories each and not so much with 3 stories each.