New statistic: Notes per hit

[bur]

Especially for those who had no hit so far it'd be interesting to know how many notes they probably need to enter for the first hit. This can be simply done by dividing the number of notes by the number of interesting hits. In the moment it's about 206 notes per hit.

Of course this isn't that accurate because it doesn't take into account the denominations entered or the country of origin. For example if you only enter 500€ bills from Malta you'll probably need more than 206 of those for a hit...

The statistic could be refined though by showing the notes-to-hit ratio for the individual countries and denominations. It should even be possible to calculate this number individually for every user by using the notes entered before for the calculation.

There are many users who enter a few notes, get no hit and then never come back. If every user had a kind of progress bar that counted how many more notes on average were needed to get a hit that would probably make some people stick to the site and keep entering notes.


I think the needed calculations are fairly easy to be done and implemented and I think that would make a great feature.

[Crazy Bob]

Well, for the most part, this feature has already been implemented. When one looks at (for example) the profile of The Netherlands, you see that the hit ratio of The Netherlands is 1:186.13 (1 in every 186.13 notes is a hit).

[Nerzhul]

On the other hand, please note that the figure in the profile resembles the hit ratio over all notes which were ever entered. For countries this usually resembles a 6+ year period taking into account the early years when the hit ratio was much worse.

A better approximation of the current hit ratio would be notes entered this year / hits this year.

For example Germany has a hit ratio of ~187 this year (321 overal) which will hopefully get even better with time.

[Dioniz]

There is also another point: statistics are taking into account both - active and passive hits. That means every hit-note is calculated into hit ratio of two or even more user's profiles.

[bur]

I didn't know this statistic was already available, good to know though.

That both active and passive hits are taken into account is okay, because as a user you normally don't care whether the hit was active or passive as long as it is a hit.

What about the idea to implement a counter on the user's start page that shows how many notes more he probably has to enter for the next (or first) hit? This number could be based on the user's country notes/hits ratio for the last year. I think that would help keep new users motivated. Currently they mostly enter about 20 notes and then never come back - if they were working for a clear goal they might stick around.

[Dioniz]

That both active and passive hits are taken into account is okay, because as a user you normally don't care whether the hit was active or passive as long as it is a hit.

OK, I can agree with that.

But regarding country and overall hit ratio it should be stated that calculations are taking into account number of entries not number of actual (different) notes. Second option would involve subtraction of number of hits from the sum of all entries.

[doiknow]

That's a hard job as hits seem tobe a highly stochastical thing. I once had a break between two hits which was at about 3 times normal hitratio. The second problem I face are the various places: Most trackers seem to track at different places thus needing different calculations which need to be merged in some kind of expected value...

[Nerzhul]

That's a hard job as hits seem tobe a highly stochastical thing.

So true. A hitratio of 1:200 doesn't mean that you get a hit every 200 notes. It's like rolling a dice ... you have a 1 in 6 chance for every number, but you don't get a specific number guaranteed every 6 rolls.

For the record, if you enter 200 notes with a 1:200 hit ratio your chance to get a hit is ~63%. For a 99% chance to get a hit under these conditions you'd have to enter ~919 notes. (Of course all these calculations are only taking active hits into account.)

More importantly, a note you already entered which turned out NOT to be a hit doesn't count towards the numbers for future active hits. So let's assume you got 919 notes from a bank and already entered 300 without getting a hit. Bad luck for you, you'll have to enter another 919 notes to get a hit with 99% probability...

This makes it next to impossible to provide meaningful suggestions like "come on, only 200 notes to go for your first hit".

Calculations taking into account passive hits are much more complicated. You'd have to combine a statistical model for note movement with a statistical model for decay. I'd rather not go there...

[androl]

When one looks at (for example) the profile of The Netherlands, you see that the hit ratio of The Netherlands is 1:186.13 (1 in every 186.13 notes is a hit).

and that is not true! If you look at the list of hits with the Netherlands, you will see that out of 100 hits, about 65 are national hits, making these 100 hits 165 note entries in the Netherlands. You'll have to divide the 9 046 537 notes by the number of note entries that are part of the 48 759 hits, which will give perhaps a hit ratio of 1:112

EBT-wide, the hit ratio is not 63 880 330 / 311 111 = 205
but 63 880 330 / (622 222 plus more for triple hits etc) ~= 100

(where can we see the number of triple and quadruple hits? would that be a new feature? )

[R/J]

(where can we see the number of triple and quadruple hits? would that be a new feature? )

That is already there:
Click Hits, then Distributions (or Verteilungen if you're using the German site). It's there.