Topic: Bitcointalk Merit Dashboard - page 31. (Read 39605 times)

Thanks. So it's not possible to check if the personal legendary activity limit for a Hero has already been reached, if the Hero don't have 1000 Merits. (If the Hero's current activity is somewhere between 775 and 1030.)

Highly appreciate dedication of  DdmrDdmr to create such informative topic on merit distribution in the forum.
Nevertheless, it might be better if you can give additional analysis on merit distribution on different member rank, for example mean +/- standard deviation, and median (interquartile range) of merits (received, sent).
With unnormal distributions, median and interquartile range should be used instead of mean and standard deviations. Presenting uncorrect statistics might  lead to misunderstandings, mis-interpretation of merit distributions on different groups of member in the forum, and should be avoided.
Additionally, outliers should be stated clearly and truncation approach should be used in some specific cases to present more correct statistics for the whole sample. (For instance, cases like nullius, iasenko, etc. should be truncated  ). After using truncation approach for outliers, mean and median of sample will be almost the same.

Yes, I used the lower Activity limit as you’ve seen. On the dashboard, the Ranking-up pipeline shows us what you’ve asked. If you click on the top left chart on the 247 number associated to "1. Could have ranked up to legendary, lacking merit" item then the user list on the right gives you their user names. From there you can click on any user and follow the "go to user profile" to see the bitcointalk user profile.
The bottom graph tells us how far these users are from their objective broken down by groups of 20%.

How do you determine if a Hero has reached the activity points needed to rank up?
If I understand well, it's not possible to determine if a Hero is not ranking up only because of the lack of Merits or because of the lack of both activity and the Merits, if the Hero has already reached the 775 activity points.

EDIT:
just checked the other topic, and saw that you use the 775 activity point to calculate with for every Hero

Is it possible to check somewhere that a Hero has reached his/her personal limit of the activity points? (and ony waiting for the Merits to come?)

Merit Dashboard

Edit 15/06/2018:

Updated Dashboard to reflect the most recent sMerit available data:
Total sMerit: 171.654    
Total Txs:       77.387
From Users:  14.796
To Users:      16.583
minDate:      2018-01-24 22:12:21   
maxDate:     2018-06-15 02:42:15


The dashboard gives you access to anyone’s complete merit history in the TX tab, surpassing the 120 day limit.

Added Tabs this week: User Summary, Section/Subsection, Rankings and WordCloud.

Thankyou zentdex for spending some time using the Merit Dashboard and Network. As you say, the dynamics of the tools bare the true value, going from a statical views to dynamical view on the data. The filter features allow us to focus on specific sets of data and derive potential useful information from it, being seeing our own personal records just a minor part to it all. Export to pdf allows us to capture snapshots of data and keep it as a reference for comparison reasons overtime.

My initial idea was to keep building on it, adding features from previous analysis into it.

I don’t believe we’ll find many forums with this level of transparency on some of their key operative features, and even less with public dashboards built around them. Well, we could say that Facebook surpasses this level of transparency without “meaning to” …

Freaking awesome job! And now I see stats that are so dynamic that enable other users to interact with it.
I even spent almost half an hour exploring, filtering, searching, sorting and extracting meaningful information.
Those cool merit network diagrams complete it all. I have fun dragging around the nodes. Thanks for creating this.

Merit Dashboard

Edit 08/06/2018:

Updated Dashboard to reflect the most recent sMerit available data:
Total sMerit: 167.409       
Total Txs:      75.332
From Users:  14.548
To Users:      16.303
minDate:      2018-01-24 22:12:21   
maxDate:     2018-06-08 02:48:03.000

Link:  BitcoinTalk Merit Dashboard   

Yes, Thursday is a bit of a bugger as it biases the data as you say. Perhaps the ideal solution would be not to pre-filter data out (so as not to lose the total aggregates), but to include some filters that allows us to narrow down the data (for example, a date interval). That should deal with both of your suggestions in one go.

The dashboard, although everything is based on the same set of data, has some external pre-rolled-up data on some tabs (Global Summary Tab is one of them). That makes it quicker for the dashboard, but limits filtering options. That means I’ll have to change the way things are done internally in order to be able to introduce granular filters by dates, going against the basic merit TXs, and not a rolled-up derived set like I do now.

I’ll add it to my backlog of tasks to do on the dashboard…

EDIT:

Just broke the "thursday is the best day to post" mith!:

I've added a date calendar filter that can now be applied on the first tab (I've a minor issue still with the time which I have to reset to 00:00:00 on the first date of the selection range). If we filter from the 29/01/2018 00:00:00 .. 27/05/2018 23:59:59 (complete weeks, skipping the first and last partial weeks) we now get:

Monday: 15,15%
Tuesday: 15,47%
Wednesday: 15,68%
Thursday: 15,31%
Friday: 14,02%
Saturday: 12,15%
Sunday: 12,21%

We can now see that Thursday is not a thrasher at all ...

Great job as usual.

Thursday is sticking out because the merit system was introduced on a Thursday and we sent a gazillion merits on that first day.

Maybe the day-of-week stats should consider only full weeks? E.g. start on the first Sunday (or Monday) after the system was introduced and also exclude the last incomplete week.

Cheers. I corrected the typo you spotted. The Merit Dashboard uses the same dataset as the merit network so as to stay synchronized between the two applications, both surpassing the 120 day TX limit of our user profile on the forum.

I’ve still got a few tweaks to do on the current tabs. For example, I’m now allowing to click on one of the graphs on tabs where user names are detailed, in order to filter the username list in a seamless manner (click again to undo).
I’ve done that so far on the Ranked-up and Ranking-up Pipeline Tabs, and need to tweak something alike on the Received and Sent Merit tabs.

Nice work, it allows us to check the merit history in detail without 120 days limitation.  I see "Set Merit" in the tab, maybe a typo of "Sent Merit"?

Perhaps... let’s see how thing playout and evolve...
 

coinlocket$, I added the hyperlink to the user profile from where you suggested, and also in the TX Tab. When you click on the user, a small pop-up menu appears and allows you to click the hyperlink to go to the Bitcointalk profile (or message in the TX tab).

This is another of those features that on the development environment works one way, but on exploitation requires a different solution (in development I can just click and go to the profile, with no prior menu selection). It not that much of an issue so I’m happy with what I’ve added.

Amazing this is the final thread for the merit, but I have a request 

Is possible to set hyperlinks here?

Maybe you and I can enter into a data transfer service agreement, once I've completed enhancements to BPIP.

I have the data, but feel no need to make all these pretty graphs when you have already done an awesome job!

The Merit Network is initially a graphical representation of all the Forum Users that one has interacted with either sending them sMerit, or receiving it from them. You can parametrize the user you want to query about. It basically shows the merit log information that is found on the forum page, but in a nicer manner and without the 120 day historical limit.

The limitation is that it needs to be updated by me, and that is not a short process. Also, it cannot be done with any granularity of less than a week, since official merit transaction file is released weekly.
I’ve recently enhanced it to include the related numerical values as well as the complete overall Merit Transaction file.

The 120 sMerits I’ve awarded to other users come from 240 Merits I’ve been awarded by other forum members for certain posts (a bit more now)...

This is a commendable effort you put in.

What is the merit network all about? How do you get over 120 merits you gave to persons as outlined in your link post, https://fusiontables.google.com/DataSource?docid=1NLEPIaT4K4pcBf4pE3tiH4w_3zoZFUIWy7bEfn83#rows:id=1

The information you presented is very upbuilding.

The actual dashboard should show dates and numbers based on your locale. It's just on the screenshots that show up based on mine. I've just set another device to english UK and english USA and it does the change correctly...

Nice stats, would be better if you used English (Google fault) rather than Spanish but it isn't that hard to figure it out tho.
+3 virtual smerit, ok <3 vod

Merit Dashboard

Edit 05/10/2018:

Link:  BitcoinTalk Merit Dashboard  


Updated Dashboard to reflect the most recent sMerit available data:

Total sMerit:  241.323
Total Txs:      112.275
From Users:    17.784
To Users:        22.543
minDate:        2018-01-24 22:12:21  
maxDate:       2018-10-05 02:51:53

Edit 04/07/2018:
Added a Tab called Reciprocal sMerit which, if used properly, could help to detect Merit Abuse (see my post on this thread: https://bitcointalksearch.org/topic/m.41487786). The Dashboard has now grown now quite a bit since the initial version...

The dashboard gives you access to anyone’s complete merit history in the TX tab, surpassing the 120 day limit.
Added this week: Post summary (distribution of sMerit per TX and TXs per post).
Added Recently: De-ranked Users, Potential Merit Sources, User Summary, Section/Subsection, Rankings and WordCloud.
-------------------------------------------------------------------------------------------------------------------------------


I’ve been considering updating the data of some of my previous analysis. Since the process is rather manual, I thought it would be nice to bring it together into a common Merit Dashboard, in order to keep it all in one place, and to ease the updating process on a more regular basis if necessary. I’ve started off by including what I’ve considered the most relevant information, using a set of tabs to hold the information for a given conceptual vision of the data.
 
The resulting dashboard is a work in progress, since I will add more features and datasets whenever possible. To mention that I do prefer the colours on my previous analytical posts (black background on the graphs), but for now the dashboard will have a white background since I’m concentrated on content and a little bit less on aesthetics.

The dashboard requires no installation, but it does result to java script (so it must be enabled in order for it to work).

Note: Screen shots show up with certain date information based on my local settings. If this works correctly, the actual Merit Dashboard should work based on your local setings... (dates and number formats, not captions).

The dashboard has the following Tabs:


1.   Global Summary:




Shows a summary of how merit is awarded, broken down by:

-   Week (Monday to Sunday- on previous analysis I may have used Sunday to Saturday since it’s the default in the software I use for data crunching).
-   Day of the Week.
-   Hour (UTC – This is not too important, but some merit awarding varies by date depending for example if the data is UTC, UTC-6 or UTC+6).
-   Weekly Ratios I’ve used before, including total aggregates at the top of the table.

Since the data originated in the public merit.txt file is published on Friday mornings, the data to the most recent week is partial (lacking most of Friday, as well as Saturday and Sunday). The rest of the weeks are complete.

We can see a clear weekly downtrend again since the end of April onwards after a light recovery in the last fortnight of April.

We already knew that Thursday is the most merited day and that UTC 8-20h are the most merited hours of the day.

New weekly users that receive Merit are around the 25%-27% area (so every week roughly one quarter of merited users are merited for the first time -> see %ToNew in the Weekly Ratios).



2.   From/To Ranks



Shows which ranks merit which ranks both in absolute value (number of awarded merits) and relative values (% of awarded merits per “from rank” to each “to rank”).

This is similar to Which Ranks send sMerit to which Ranks - and who ranked up, but now updated.

Comparing current data to the analysis above stated, we can see for example that:
-   Legendaries keep roughly the same awarding proportions overall as they had around end of march, being the most favourable movement increase for Jr. Members (4,37% -> 6,18%).
-   Heroes keep nearly the exact proportions as end of March.
-   Sr. Member have dropped their awarding rate to Sr. Members (23,15% -> 20,56%) and have moved the drop to Members (23,02%->24,46%) and Jr. Members (6,57%-> 7,28%).
-   Full members have dropped a bit on their ratio of awarding full members (31,74%->29,52%), moving the difference to Members (26,69% -> 29,57%).
-   Members awarding to Full Members has moved from 20,26% to 17,15%,  being Members the most favoured increase (42,92%-> 47,07%).
-   Jr. Members now favour less Full Members than before (19,48% -> 14,01%), but increase the Newbies rank (4,22% -> 8,62%).

All in all there are slight changes, but not too significant.

Note that the rank is derived from the current user profile, and not at the time of actual awarding which may have been different (we cannot access the rank at time of awarding, so this is the next best thing we can use).


3.   Received Merit



Shows the overall received Merit breaking it down by the Ranks that send it, month, specific date and user names along with the aggregate quantity they’ve sent.
By default, the data shown is the overall merit, but you can filter by user name/s to delimit the view to a specific user set  (i.e. filter by your own username to see the merit you received).
 

4.   Sent Merit



Shows the overall sent Merit breaking it down by the Ranks the merit has been sent to, month, specific date and user names along with the aggregate quantity they’ve received.
By default, the data shown is the overall merit, but you can filter by user name/s to delimit the view to a specific user set (i.e. filter by your own username to see the merit you sent).


5.   Ranked-up



This gives us the amount of users that have ranked-up using the new merit system, indicating their starting rank and current rank. This is similar to what I’ve done before numerically in posts. To see a specific set it is easier to filter by (probable) initial rank and current rank.

I’ve omitted including combinations relative to ranks that are not amongst the conventional ranks that need merit to rank-up. There are even some de-ranked cases (omitted), probably caused by purchasing an account and deleting the posts.


6. Ranking-up pipeline



This gives us a global vision of how many users are depending on what (activity, merit, both) in order to rank-up. This only considers users that have been awarded Merit (in at least 1 Merit TX), not those that had a Merit airdrop and have not received any further Merit.

This is similar to Stats on the Rank pipeline - How many are we on the way to ranking up? extension

Comparing the above mentioned analysis (created during mid-April 2018) to current dashboard data, we can see that:
-   We’ve gone from 135 to 219 Heroes that could have ranked-up, but lack Merit.
-   We’ve gone from 297 to 444 Sr. Members that could have ranked-up, but lack Merit.
-   We’ve gone from 1.146 to 1.919 Full Members have not ranked-up due to lack of Merit (by contrast, those lacking both activity and merit have decreased 1.458 -> 911).
-   We’ve gone from 2.032 to 2.711 Members that could have ranked-up, but lack Merit.
-   We’ve gone from 1.914 to 2.460 Members that could have ranked-up, but lack Merit.

In summary, merit is slowing down the ranking process (as we already knew), and candidates awaiting “just” to earn merits in order to rank-up are greater by number.


7.   TXs.




A list of all the Merit Transactions. The idea here is to filter by User From Name or User To name to see all Tx history for a given user/s. This can also be done on the Merit Network.


8.   Merit Network



This should contain and embedded link to Our very own sMerit Network Picture - enhanced with access to all sMerit TXs.

Unfortunately, I can see the embedded webpage fine in development, but not once I view the published Merit Dashboard. Until I find a solution (if there is one, which I’m not too positive about), I’ve placed a button that takes us there.


Dashboard Usage:



Using the dashboard is pretty straight forward. It may be good to mention that:

If you scroll down you’ll find a very useful toolbar that allows you to:
-   (<-) Undo if you apply a filter to the data and cannot revert easily to the previous unfiltered situation.
-   (-> )Redo.
-   (rectangle with a downward arrow) Save as image or pdf.
-   (rectangle) Full Screen.
-   (ESC) Exit Full Screen.

Also, when applying a filter (for example in the Received Merit Tab), you can type in the name and press the plus sign to add the name to the filter (multiple names are allowed), or use the magnifier to search for all name that contain the typed string (and then select the desired names).
To delete a filter you can hover over the filter value until a cross sign appears to delete the item, or use the undo arrow on the toolbar mentioned above.