profile/splitti3
Summary
Split the first table of data in a CGATS format file, into two files,
choosing the sets randomly. Usually the CGATS file will be a .ti3
format file, and the intent is to create a test and reference data set,
to verify the operation of colprof.
Usage Summary
usage: splitti3
[-options] input.ti3 output1.ti3 output2.ti3
-v
Verbose - print each patch value
-n
no Put no sets in
first file, and balance in second file.
-p
percent Put percentage% in first file,
and balance in second file. (def. 50%)
-r
seed Use
given random seed.
input.ti3 File to
be split up.
output1.ti3 First output file
output2.ti3 Second output file
Usage Details and Discussion
splitti3 together with profcheck
provides a way of verifying how well the profiling process is working.
By splitting up a .ti3 test
set into two parts, one that will be the data file that the profile is
created from, and the other being the independent set used for
verification, some measure can made of how well a devices underlying
behaviour is being modelled by colprof. Only the first table is
transferred from input to output.
The -v flag prints out extra information during operation..
The -n parameter specifies the split between the first and
second output files as a given number of sets of data for the first
table.
The -p parameter specifies the split between the first and
second output files as a percentage in favour of the first table.
The -r parameter provides a
way of making the random split reproducible, by allowing the random
number seed to be specified.
Typically a large test set might be partitioned into two files, the
first used to generate a profile, and then the second used with profcheck to check how well the
profile patches the balance of the test set points.