How to improve Gnucash QIF import matchings by adding account to each transaction

Question

I'm trying to suppress the longest part of my QIF import process which is manual matching of each transaction to each account.

I have the bayesian matching activated but it performs poorly, matching maybe 10% of all transactions. I think it is because it is based on an exact string match between the transaction name and the account memo field, and typically the transaction name will have more infos (like the date, etc) -- not sure though.

At http://wiki.gnucash.org/wiki/FAQ :

An alternative approach to creating the account tree in Gnucash first is to include your categories and accounts in Quicken's QIF export. This option is available in many versions of Quicken. If you have this data in the QIF file, Gnucash will create your accounts automatically.

So I wrote up a minimal QIF file :

!Type:Bank
D05/02/2015
T-40.00
PRETRAIT DAB 01/02/2015 CIC PARIS
^

Say I have an account named Cash in Wallet (with parent account: Expenses), and I want to assign the transaction to this account.

From that source, it seems that the account name must be entered on its own line before the end marker ^, like so :

!Type:Bank
D05/02/2015
T-40.00
PRETRAIT DAB 01/02/2015 CIC PARIS
Cash in Wallet
^

(I tried with entering the absolute name Expenses/Cash in Wallet, too)

When importing that file, the match still not occurs :

What edit should I do on the QIF file so the import assigns existing Cash in Wallet account as Gnucash account name to the transaction ?

Answer 1

Everything is clear now that I found QIF specification.

I forgot the leading L that indicates it is a category field.

!Type:Bank
D05/02/2015
T-40.00
PRETRAIT DAB 01/02/2015 CIC PARIS
LCash in Wallet
^

does work.

If someone is interested I coded a small tool to help automate the categorization process, available at https://github.com/Kraymer/qifqif