advertisement: compare things at compare-stuff.com!
next up previous contents
Next: Prediction of other architectures Up: Prediction of mainly- architectures Previous: Ribbon, barrel or sandwich?   Contents

Non-ribbon architectures

Given that a mainly-$\beta $ class domain does not have a ribbon architecture (this can be predicted quite well, see above), how well can its architecture be predicted? Only 48% of the 107 non-ribbon domains are correctly predicted (31% random). Again, the barrel and sandwich architectures give the best Matthews coefficients, but they are both less than 0.35. The reliability quartiles are only 54%, 63%, 52% and 22%. Predictions with such poor performance can only give a suggestion of the likely architecture. Further steps in the hierarchical approach for mainly-$\beta $ domains would currently be unwise because the numbers are becoming small and the compositional differences are becoming weaker.


Copyright Bob MacCallum - DISCLAIMER: this was written in 1997 and may contain out-of-date information.