⭐ Merited by coinlocket$ (1)
We have noticed before that some users send 50 merits to a nonsense post and I think some of them are now very careful in sending their sMerits which is a good thing. They are somehow forced to find a bit of quality in a post. Your efforts are appreciated. Keep up the good work.
Thank you. You are right, the number of member who give 50 merits in a single transaction decreases.
Seems obvious to me that this doesn't represent distinct entities but rather just new accounts--multiple accounts of people.
Nice data presentation, OP.
Nice data presentation, OP.
That is also what comes in my mind when I have see the values. But just like you said, it will be hard to determine which are alt accounts even considering IP. Thank you.
What activity number are you using for heros to rank up?
775 or 1030 or somewhere in between?
775 or 1030 or somewhere in between?
I use 1030, because It's really hard to pinpoint which is which.
some of them are afk/don't care
I'm with you by that. There are estimated more than 1,000 accounts who received merits but did not give back enough to others.
I would be interested to see how long their posts are compared to others (specifically those who are one rank higher), where these accounts tend to post (what sections), if these people are participating in any kind of advertising campaigns, how much trust these people have sent/received.
Good suggestion, thank you for that. It could took me a longer time to do that and may require a greater computing power.
So, why not publicly publish datasets?
~snip
Because the dataset is not what other user expected to be. I could be more harder to understand. For example, I have a separate table for user id and name, another for merit, another for post, activity and rank, another one for dates, table for who send and who received among others. I then create a queries to see what I am looking. Is it possible to look at the average amount of merit given, per transaction? I'd like to see if there's a correlation between rank and the people who are gifting merit to friends or alt accounts.
Parameters likely need to include:
Parameters likely need to include:
- Average of merit given per transaction, by rank level
- Being able to look at that average over shorter time frames would be useful (a week or 14 days maybe), the longer the duration the more likely the 1 merit transactions will dull the results, we're looking for spikes
- Can we isolate transactions by forum board? Some boards are more likely to have these questionable transactions (e.g., alternative currencies, local, etc.)
- We should isolate the received side of the transaction, an alt account is more likely to receive transactions that are arbitrary gifts than due to genuine quality of their posts - obviously just an unsupported hunch on my part
- Exclude people who've never given any merit
- Average amount of merit given, per transaction? - 2.16 with deviation of 3.97
- Average of merit given per transaction, by rank level - see table above
- Can we isolate transactions by forum board? Some boards are more likely to have these questionable transactions (e.g., alternative currencies, local, etc.) - It is possible, but what is the purpose of isolating those boards?
- We should isolate the received side of the transaction, an alt account is more likely to receive transactions that are arbitrary gifts than due to genuine quality of their posts - obviously just an unsupported hunch on my part -I have stat that dived those two. see II. Giver and Receiver. But I can't isolate those who are alts because I have no further info who among are alts
- Exclude people who've never given any merit - It is already excluded
Thank you for your suggestions.
I completely appreciate this STATS from the OP. I hope that some important statistics from this forum will show up lately.
Thank you for your appreciation.
~snip
Thank you for appreciating my posts.
I guess it is because of users do not bother to read tons of remaining comments to the thread, it is actually true, well, one more idea for a new graphic 
Nice suggestion, if the page view per user data is available that could be done.