Have you tried leaving off the format option, letting it write in the default 
format?  Or specifying binary, since metric-information hints at a preference?


On Oct 29, 2012, at 7:10 AM, Colin Reveley <[email protected]> wrote:

> Hi - 
> 
> I'm interested in correlating two metrics.
> 
> for the sake of argument these are the Shape index and integrated folding 
> index of a surface.
> 
> Rather than treat the rows as locations in space (nodes) and the columns as 
> observations on those locations, I'd like to treat each row as as an 
> observation, and each column as a location in space instead.
> 
> So I want to transpose the two metric files. run stats. transpose the results 
> back so I can plot them on the surface. right?
> 
> The trouble is that when I do that I get:
> 
> $ caret_command -metric-information R.SHAPE.metric 
> Filename: R.SHAPE.metric
> Number of Nodes: 70962
> Number of Columns: 1
> Column      Minimum      Maximum           Mean     Sample Dev     % Positive 
>     % Negative   Column Name
>      1       -2.550        6.000          1.001          0.505         97.696 
>          2.304   MASKED_SHAPE_INDEX_norm
> 
> 
> $ caret_command -metric-transpose R.SHAPE.metric R.SHAPE_transposed.metric 
> -WRITE-FILE-FORMAT-METRIC XML_BASE64_GZIP
> Input Filename: R.SHAPE.metric
> Output Filename: R.SHAPE_transposed.metric
> Number of Nodes: 70962
> Number of Columns: 1
> 
> $ caret_command -metric-information R.SHAPE_transposed.metric 
> 
> METRIC FILE INFORMATION ERROR: Error  : Decompression of Binary data failed.
> Uncompressed 0 bytes but should be 4 bytes.
> 
> 
> and there seems little I can do to fix this. I've tried a bunch of things.
> 
> clearly I need to do something to the metric file before I run the transpose 
> command. not sure what though.
> 
> thanks,
> 
> Colin
> _______________________________________________
> caret-users mailing list
> [email protected]
> http://brainvis.wustl.edu/mailman/listinfo/caret-users


_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users

Reply via email to