I'll send you further detail and a sample training file and script
today or tomorrow, but the long and short so far is that Text columns
are still being ignored (changed to '?'s) when running a script which
references the training file containing the text column. It happens
whether the text is double-quoted or not.
Also, the script will no longer run by specifying it on the command
line, even if the command line is run directly from the EasyNN folder
and there are no spaces in the script filename, as in
, run from \Program files\Easynn-plus, with test.scr in that folder.
Again, the script runs from the File Run Script File menuitem ok, in
the sense of pulling in the referenced training file.
Also, this isn't really a bug, but the script doesn't seem to continue
when there are minor problems in the training file - specifically,
when there are duplicate rows. Is there a command to place in the
script, that lets EasyNN know to automatically include duplicate rows?
I'll experiment further this evening, try to narrow down the apparent
problems, and get some files to you.
Two feedback comments:
1) Include more sample scripts, that perform various aspects of the
NN creation and processing. Perhaps you can include a script for each
of the sample projects packaged with EasyNN.
2) In some future version, put in a Generate Script menuitem, which
takes the current network and settings, and creates a partial script
which can be cloned, modified, and used externally. The generated
script would contain ctm() references to all of the columns in the
current network, and contain script commands analogous to the current