I wonder if I am missing something - I ran waffles_learn.exe crossvalidate and
the only results are the scores. Is it possible to save the models that are
trained to stdout as well? Or do I need to call "train" with specific split
files as inputs? It seems like maybe crossvalidate was supposed to save me
from having to create all the splits etc.
thx this looks like a pretty handy tool.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yes, "crossvalidate" was designed to be a high level tool for easy
convenience. If you want more control, you can do it at a lower level with a
script that calls these tools:
waffles_transform shuffle
waffles_transform splitfold
waffles_learn train
I prefer to do it at a low level with a script because then I can plug in
other tools as needed. For example, you could use "waffles_plot model" to
generate a visualization of each model, or you could use "waffles_transform
mergevert" to gather accuracy results into a single data file, and then use
"waffles_plot scatter" to plot a chart.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I wonder if I am missing something - I ran waffles_learn.exe crossvalidate and
the only results are the scores. Is it possible to save the models that are
trained to stdout as well? Or do I need to call "train" with specific split
files as inputs? It seems like maybe crossvalidate was supposed to save me
from having to create all the splits etc.
thx this looks like a pretty handy tool.
Yes, "crossvalidate" was designed to be a high level tool for easy
convenience. If you want more control, you can do it at a lower level with a
script that calls these tools:
waffles_transform shuffle
waffles_transform splitfold
waffles_learn train
I prefer to do it at a low level with a script because then I can plug in
other tools as needed. For example, you could use "waffles_plot model" to
generate a visualization of each model, or you could use "waffles_transform
mergevert" to gather accuracy results into a single data file, and then use
"waffles_plot scatter" to plot a chart.