« first day (1576 days earlier)      last day (757 days later) » 

07:09
Sunday!
 
7 hours later…
14:24
@Martin if you're around, I don't understand why rouding my log10 of text lengths get these weird jumps: data.stackexchange.com/math/query/1674211?decimals=3#graph it seems to lessen if I round to 4 decimals but there is either something wrong with how I think I can use that round function or there is a strange side-effect with taking the log. I'm trying to get a kind of "distribution" of post lenght.
 
2 hours later…
15:59
I do not have a quick fix, but I will just point out things which might influence this. (Although it took me some time to realize.)
I will try with decimals=2.
The value round(log10(len(text))=1.72 means that we're counting posts with lengths 52 and 53 - and the values 1.71 and 1.73 each have only one corresponding integer value.
So under the (reasonable) assumption that the number of posts with each lenght close to 50 is approximately the same, the value for 1.72 is going to be twice as large as the neighboring values.
Hello, my apologies for jumping into this discussion
Could someone kindly tell me what acceptedanswerid means in SE database?
I did go over here - meta.stackexchange.com/questions/242739/…, but there couldn't find any documentation about it
Thank you!
@desert_ranger The id of the accepted answer.
I.e., the one with the "green tick" - if there is such answer.
@Martin I see. So, if I were to look for questions with accepted answers, I'd want to look for rows with an "acceptedanswerid" ?
I would expect AcceptedAnserId IS NOT NULL.
@Martin Thank you so much! That's all I wanted to know
Have a great day :)
16:09
@desert_ranger You can see an example of such query here: How to use Data Explorer to find my questions for which I have not accepted an answer?
And rene actually created a network-wide query: How can I view all my unaccepted questions across SE?
@Martin Thank you once again!
I wonder whether it would help to round down - I see that originally you wanted to use floor function.
16:46
A few experiments suggest that this doesn't really help. (It just moves jumps to a different place.)
 
3 hours later…
19:47
@Martin yes, that makes sense. I assumed using the log10 would help me here as there would be many posts with less than 1000 chars while I am trying to analyze post size at the tail end of that graph. Your queries at least shows/explains to me the behavior, so that is helpful. Thanks.
So this then shows how quick the number of posts drop for increasing post length in a log10 scale on the x-axis. data.stackexchange.com/math/query/1674455?decimals=1#graph
this is less useful, right? data.stackexchange.com/math/query/1674465?decimals=1#graph that is just plain len(text) against count

« first day (1576 days earlier)      last day (757 days later) »