Data Explorer (SEDE)

« first day (31 days earlier) ← previous day next day → last day (2060 days later) »

3:35 PM

Q: A statistical question about statistical questions

In Mathematics Stack Exchange, has there been an increase (since the end of March) in the frequency of questions about probability and statistics (say, as a fraction of the weekly total of questions asked)? I ask because I hardly visited MSE from April to July, and when I returned in August I se...

SEDE seems like a natural tool here.

I have tried to create some queries to check the numbers by month.

Here is what I tried to get the relevant numbers: data.stackexchange.com/math/query/891708/…

By searching on the site for questions in one month and restricted to statistics tag, at least the first number seems to be correct.

And in this one I tried to create also a graph: data.stackexchange.com/math/query/891711/…

A natural tool to check something like this is Data Explorer. I have tried to create a very rudimentary query in the Data Explorer chatroom. I hope somebody with better knowledge of SQL will suggest something which works better. — Martin Sleziak 11 secs ago

rene

4:11 PM

It is not bad at all, I wouldn't be so pessimistic about your SQL Skills. Here is one with the EOMonth function so your graph gets dates on the x-axis: data.stackexchange.com/math/query/891716/…

Martin

I suppose that if I want breakdown by week, then I can no longer use formator something similar and I have to try something different: data.stackexchange.com/math/query/891718/…

rene

yeah, that will work.

Martin

BTW when SEDE also offers a graph. Does the query need to have only two columns? Does the first column need to have some special values?

I have used ROW_NUMBER since I thought that to get the graph I need the first column in the results be increasing integers. (Or at least I remember seeing some queries that had a graph and the first column of results was like this.)

rene

the first column either needs to be a date or a number and its values go on x-axis, all other columns go as Y-values

STR(Year(CreationDate))+'-'+STR(DatePart(week,CreationDate)) will break for weeks < 10

Martin

:-(

rene

4:18 PM

it is not 0 padded

20181

201810

let me see

@Martin this will work: data.stackexchange.com/math/query/891726/… notice how I use the group by and order by. You don't need the string juggling there. In the projection implemented a simple right padding trick.

Martin

I see. Somehow I had fixed the idea that I want to use in the group by the same thing which I have as the output in results.

The way to display year-week in your query is simpler. (IIRC I have copied this stuff from some other query which used Month and now I just played with that to get from that in some way week rather than month.)

BTW it seems that the OP checked the query and plans to return to this back at some point. (Judging by the recent comment on the linked post.)

Thanks a lot for the help!

rene

Martin

5:27 PM

Just for completeness, I'll add also a version of the last query (per week) which only shows percentages and contains a graph: Percentage of questions in a given tag per week

« first day (31 days earlier) ← previous day next day → last day (2060 days later) »

all rooms