Haojin Zhai's phylogenetic Tree data
haojin_zhai@hotmail.com

These were used in the analyses in Haojin's dissertation:
Zhai, H. (2016). Principal component analysis in phylogenetic tree space (Doctoral dissertation, The University of North Carolina at Chapel Hill).
https://search.proquest.com/openview/0f690369abb088a36393b79935103090/1?pq-origsite=gscholar&cbl=18750

These came from finding artficial landmarks for Liz Bullitt's MRA derved Bran Artery Trees.  
This data set had 109 trees, with corresponding age, gended and handedness (not very useful, since almost all were right).

Key idea to generate phylogenetic trees was Ipek Oguz's corresponding landmarks usd as an artifcal leaf set.  

An early analysis used the Cates Whittaker software.  This lost some cases for which the software failed.
These should be in the Matlab data file:
DataSet4_New.mat
which features: 
85 successes in processing.  
The successful case numbers are in:
    TreeDataCase
Note:  must use these case numbers to get correct ages and genders


A later version used Ipek's disseration software, which should give better landmarks, but fewer trees, as there were more software failures.
These should be in the Matlab data file:
DataSet5_New.mat
which features: 
67 successes in processing.  
The successful case numbers are in:
    TreeDataCase
Note:  must use these case numbers to get correct ages and genders


An interesting question (as yet not followed up) is which gives better statistical inference (e.g. age or gender effects).  The first set is larger (more data points), but is likely of lower quality.