TODO list for Perl module DateTime::Even::Predict

12/22/2009
* Can we combine the interval and distinct buckets into one full bucket list? The names should mean there's no
  collisions (just confusion, maybe too much) and the 'type' identifier says how to use it. Probably not a
  good idea.
* train() needs to reset the bucket values each time it's called; right now they'll just increment endlessly

12/18/2009
* We definitely need to add the interval_buckets back in. Right now there's no way
  to predict something like the Chiefs next Superbowl victory. With the years 1966
  and 1969 you get a mean of 1967.5 and a standard deviation of 1.5. We have to
  use intervals to find the next date.
* Is there a way to use import() outside of a BEGIN block so that export tags can be imported?
* Change new() and predict() so options can be globally set in the object and overridden in
  each call to predict().
* We could create a bucket for is_weekend_day and create a custom callback for the accessor.
  We'd have to do duration differently, though. 

12/17/2009
* Finish writing up pod so module can be alpha-released on CPAN
* Right now we are trimming off any date-part that is smaller than the smallest bucket we
  have turned on. We need to make it so this is done in the comparisons, rather than
  actually modifying dates we are given. OR MAYBE NOT, because of truncate()?

12/16/2009
* Due to possible DOS attack that can be done through very large duration operations (thousands
  of years in the future, etc), we'll need to add some sort of protection, I think.

??/??/????
* Add a clustering() method so that clustering can be turned on or off whenever
