One of our users, Richard Bates, had further thoughts about classification dictionaries which I'm including here as background:
Thanks for getting back. I have been editing in standard CSV mode so no real problem there. I just wondered about the ability to adjust in a more intuitive/interactive manner or better to be able to interrogate the data so that you could see at specific points in the data set what the various (broad and fine) results were where you had ground truth information. That way one could perhaps better build the right dictionary.
Can use matplotlib to generate graphs, e.g. http://matplotlib.org/examples/api/histogram_path_demo.html. Easy enough to do, but the interactive component would be more challenging and might be best performed using something beyond just GP, but it could be done in a simple way initially.
Alternatively, could make BPI predictions and intersect them with a samples file for comparative purposes
One of our users, Richard Bates, had further thoughts about classification dictionaries which I'm including here as background:
Thanks for getting back. I have been editing in standard CSV mode so no real problem there. I just wondered about the ability to adjust in a more intuitive/interactive manner or better to be able to interrogate the data so that you could see at specific points in the data set what the various (broad and fine) results were where you had ground truth information. That way one could perhaps better build the right dictionary.
Can use
matplotlibto generate graphs, e.g. http://matplotlib.org/examples/api/histogram_path_demo.html. Easy enough to do, but the interactive component would be more challenging and might be best performed using something beyond just GP, but it could be done in a simple way initially.Alternatively, could make BPI predictions and intersect them with a samples file for comparative purposes