Hacker News Reader

Comment by orless | Hacker News Reader

When applying for a loan, mobile phone contract, or even trying to rent an apartment in Germany, the Schufa score - Germany's credit rating - is decisive. If you have a few "points" too little, your application is refused. (Computer says "No" to your new smartphone or apartment.) However, the calculation of these credit scores - done by the private Schufa company - is fully intransparent. The formula is a trade secret, and as such not open to the public.

We want to change this intransparency with the project OpenSCHUFA. Open Knowledge Foundation together with AlgorithmWatch want to reconstruct the Schufa algorithm with "reverse engineering".

View on HN

How big is your dataset?

I'd expect the logic backing a legacy/old school company like this would turn out to be an expert system. If that's the case, and you have a LOT of data, you'd likely get very far by just feeding a bunch of SCHUFA applications and resulting scores into SKL's decision tree classifier.

They are easy to visualize, they likely model what's going on behind the scenes, and it takes very little effort to give it a try.

I cannot answer, I'm not affiliated with the project. Better ask @vavoida.

„...even trying to rent an apartment in Germany...“

While this MAY be true in a lot of cases it certainly is not all the time. From my experience 50/50.

see this information https://www.settle-in-berlin.com/what-is-schufa/

Not all the time probably, though in my experience, especially if it's not a flat share but a proper apartment, it's definitely closer to 100% (Berlin).

I guess the likelihood increases if you go through one of the newer screen-tenants-first-online-platforms or the building has hired a property management company. Going through newspaper-classifieds or owner-let apartments yielded 100% no Schufa for me in Munich + Berlin.

Really? Schufa was the first thing anyone asks when I was interested in a new apartment. My schufa score is abysmal, but luckily the score is not included in this document, only the negative entries.

50/50 may hold true if you are renting directly from the owner. If a real estate agent is involved (or something like degewo) it is somewhere around 100% of all cases.

+1 for owner. Luckily you can choose (still).

How do you want to reconstruct it exactly? The SCHUFA scores are only accessible for free once a year, so it would take quite some years to reconstruct them.

If you're going one at a time. You can improve this by increasing your number of datapoints (people's scores)

> For this purpose we call for data donations:

> Everybody can request their free SCHUFA-Score from Schufa under selbstauskunft.net/schufa and donate the data to our project.

How do you plan to do that? Which bits of information can you gather that allow you reverse engineer significant parts of their process?

Consumer's have a right to get an overview of all their collected data as well as the calculated (category) scores once each year for free. These could essentially serve as input output pairs.

I know what the reports contain. And I do not see how you could gather any meaningful insights from it.

It is very likely that there is relatively simple formula that maps the input data to credit scores.

Given enough inputs and outputs, you can figure out the formula.

I don't see them collecting enough relevant inputs.

We will see.

Reply on news.ycombinator.com