-
Could sub-class Pandas data-frames to add clever functionality.
- Virtual columns (accessed in the same way)
- Caching
- Lazy construction
-
Use symbolic regression to find a correlation between evolved difficulty measures and "observed" difficulty (num. evaluations, time, num. solutions found). Could we normalise between 0.0 and 1.0?