[GNC] automatic reconnaissance of a transaction

[Greg Feneis]

Just to be clear, when a user selects a file for import, it's transactions
appear in the matcher dialog. This is where the user must correct whatever
assignments are not done, or are done incorrectly.  The matcher only learns
from corrections made from within this dialog.

Once the user is past this dialog, and the transactions have been imported
and appear in a GnuCash register, if some error is found and gets
corrected, the matcher doesn't learn from this correction, so it will
likely fail the same way the next time.

Kind regards, Greg Feneis
(Pixel 3)


On Wed, Dec 4, 2019, 07:39 Joffrey Bienvenue <joffrey.bienvenue at gmail.com>
wrote:

> Thanks for the update!
>
> Yeah I can understand when description changes but here I'm referring to
> transactions that are always the same and never get automatched....
>
> I'll look at the options you gave me.
>
> Cheers,
> Joffrey
>
> On Tue, 3 Dec 2019 at 22:57, John Ralls <jralls at ceridwen.us> wrote:
>
> >
> >
> > > On Dec 3, 2019, at 4:16 PM, Joffrey Bienvenue <
> > joffrey.bienvenue at gmail.com> wrote:
> > >
> > > Hello
> > >
> > > When I import transactions from my bank accounts, some transactions,
> very
> > > few in fact, are recognized by GnuCash. It used to be better at
> > recognizing
> > > but I find it's less and less.
> > >
> > > Is there a way to help that "reconnaissance" ?
> > >
> > > Some transactions I am registering are *always* posted in the same
> > category
> > > (eg. Expenses->Groceries) but it keeps on not being recognized.
> >
> > The matcher works mostly from descriptions. Have those changed
> > significantly?
> >
> > There are two things you can do: Always update the matches and from time
> > to time use Tools>Import Match Editor to remove obsolete matches that
> might
> > lead the match algorithm astray.
> >
> > If nearly all of your transactions will match descriptions exactly with
> > their predecessors then the non-Bayesian matcher may work better, but in
> > all other cases the Bayesian one will be better in the long run, though
> it
> > requires more training at the outset to start getting good matches and
> > needs to be maintained by reviewing matches and correcting mismatches *in
> > the matcher* to keep on track.
> >
> > Regards,
> > John Ralls
> >
> >
>
> --
> --
> Je roule pour la santé des hommes. I ride for men's health
> https://gentlemansride.com
> _______________________________________________
> gnucash-user mailing list
> gnucash-user at gnucash.org
> To update your subscription preferences or to unsubscribe:
> https://lists.gnucash.org/mailman/listinfo/gnucash-user
> If you are using Nabble or Gmane, please see
> https://wiki.gnucash.org/wiki/Mailing_Lists for more information.
> -----
> Please remember to CC this list on all your replies.
> You can do this by using Reply-To-List or Reply-All.
>