Write Report

[MichaelAquilina]

How are we going to organise the report?

I think I should write about the following since I wrote most of them:

• Feature Selection
  • zipfian distribution
  • trimming very common words and very rare words
• Feature Weighting
• Text Pre-Processing
• Inverted Index Structure (Although this can probably be left out since it does not have much to do with the actual report)

I can also write about the KNN algorithm once I write it.

Other sections we deffinately need to include:

• Naive Bayes implementation
• Info on Decision Tree
• Info on Perceptron
• Results (+ what was used e.g. cross validation) including accuracy, recall etc...

Any input on this @xhochy ?

[xhochy]

• Weka has a KNN, but as a clusterer, not classifier. Slightly different interface but should be able to act as a classifier after an initial clustering
• I should write about the E-Mail Parsing
• We should make some figures/tables, basically we need the numbers first, so we should really start making some runs to get them.
• More parameter values for the Decision Tree and if possible we should state why we have choosen them (sadly we're in ML were stuff magically works)

[xhochy]

I'm going to split this up into subissues, so we can assign each to the one who is working on them

[MichaelAquilina]

Sounds like a plan!

On Wed, Dec 11, 2013 at 12:33 PM, Uwe L. Korn notifications@github.comwrote:

I'm going to split this up into subissues, so we can assign each to the one who is working on them

— Reply to this email directly or view it on GitHubhttps://github.com/KillaW0lf04/SpamFilter/issues/19#issuecomment-30316587 .

Michael Aquilina