HabroStatistics - at what time is it better to publish an article on Habré?

Astrologers have announced a month of statistics, and perhaps I will join today to the authors of statistical articles.

Surely you wonder - when is it better to publish an article? On weekends to read it carefully? On Friday, so that everyone has a good mood? On Monday, when after the weekend everyone crawls out reading Habr? Or maybe it is better to choose the time of the lunch break? At once I will make a reservation that this question is important not only for Habr - approximately the same can be said about almost any publications, if there isn’t some very harsh concrete. I have my own hypothesis, I will tell about it at the end of the article.

Of course, this question has already been asked . And answered . And draw statistics . A lot of statistics ! So why another article on this? But why? Almost anywhere is not taken into account that should be taken into account at the same time as the publication and day of the week. On Monday, Monday morning has cardinal differences from Sunday morning. Also, no one took into account the holidays. And also, many took as the date of publication of the post the date of the first comment, so as not to bother with the conversion. And if an article is published at some non-standard time when people are sleeping, then comments there may appear very soon ...

I honestly tried to find someone already prepared statistics, but alas - some were unsuitable, some of the links died, the rest were unsuitable because of the collection of obviously incorrect information. So go ahead - write your bike. I will not give the code (obviously), for whom it is interesting - it can look at github , you can also download the generated database (now about six months, then there will be more).
')
Download, raspars, poured into the database. Well done. We look. Now the truth will be revealed - when is it necessary to publish, so that there will be many views, comments, the rating grows and ponies run across the lawn. For a start, let's see when you can get the most rating.

SELECT CEIL(AVG(rate)) AS avg_rate, COUNT(id) AS quantity, CEIL(AVG(comments)) AS avg_comments, CEIL(AVG(views)) AS avg_views, WEEKDAY(created)+1 AS DAY, HOUR(created) AS HOUR FROM posts WHERE DATE_FORMAT(created, '%m-%d') NOT IN (SELECT `date` FROM holidays) AND rate > 0 AND comments > 0 GROUP BY WEEKDAY(created), HOUR(created) ORDER BY AVG(rate) DESC

Explanation: we exclude from the sampling holidays and posts that have a rating and the number of comments not more than zero.

*drumroll*

Results Plate

avg_rate	quantity	avg_comments	avg_views	day	hour
149	2	295	58,000	2	one
91	one	21	52,000	7	6
86	3	83	42334	3	four
81	four	96	49500	2	6
75	3	135	26334	five	five
68	nineteen	15	14316	one	0
67	3	80	30334	7	9
65	five	61	24,000	6	22
64	five	78	33600	four	2
63	13	56	19693	6	sixteen
55	3	thirty	24334	6	9
55	four	25	23,000	one	7
51	12	80	15,000	3	one
49	6	58	26,500	five	3
47	sixteen	40	15407	one	nineteen
44	9	35	23223	3	23
43	ten	61	16900	2	7
43	7	20	8660	7	3
42	7	27	23,000	five	one
42	ten	32	23700	6	14
41	2	ten	24500	6	four
41	12	37	18584	one	one
41	15	eleven	17934	7	22
40	6	62	24667	2	22
40	3	70	16,000	7	7
39	13	41	23154	five	0
38	five	21	17,000	6	3
38	sixteen	53	18813	7	sixteen
38	6	41	18167	five	2
37	6	39	18,500	7	2
37	ten	26	22800	3	0
36	sixteen	32	15750	7	15
36	3	26	21,000	6	five
35	five	29	12200	3	3
35	9	34	13112	7	14
35	12	17	19834	7	17
34	eight	17	15675	6	18
34	28	24	22,000	four	17
33	59	28	17323	3	eleven
33	6	28	18334	6	nineteen
33	eight	35	18542	6	13
32	14	18	18715	7	23
32	five	74	15600	7	12
32	nineteen	17	17737	four	nineteen
32	37	24	16649	3	18
32	7	24	21429	3	7
31	one	four	13,000	3	6
31	17	28	16530	four	22
31	four	nineteen	15500	6	ten
31	17	27	20236	2	20
31	12	12	8622	one	23
31	13	21	16077	one	eight
thirty	five	26	19400	five	22
thirty	eight	thirty	19125	6	0
thirty	48	21	14917	3	14
thirty	23	23	15566	2	18
thirty	41	18	16854	3	15
29	3	18	8667	7	eleven
29	eleven	67	15910	one	20
29	2	eleven	6000	6	23
29	15	33	16667	four	20
29	eight	21	16,500	7	21
29	eight	22	24750	four	3
29	five	43	21,000	five	four
29	60	sixteen	12500	one	13
29	eleven	28	17273	3	2
29	12	26	12500	3	21
29	48	23	11855	one	9
28	23	26	13044	3	nineteen
28	34	27	14824	four	ten
28	12	25	14584	four	7
28	47	28	15979	2	15
28	45	27	18489	five	13
28	26	23	13462	four	9
28	eight	14	18250	6	21
28	62	27	14203	3	12
27	6	14	13834	one	2
27	3	25	15,000	7	ten
27	58	20	13242	2	eleven
27	39	29	14231	2	sixteen
27	2	44	7500	four	five
27	47	21	13362	four	sixteen
27	39	18	15565	five	15
27	61	21	15558	one	12
27	51	22	12311	four	13
27	eleven	37	15546	five	20
27	6	84	9167	four	eight
27	nineteen	20	12632	five	17
26	eleven	25	15637	five	eight
26	22	24	13773	four	18
26	37	23	11621	one	17
26	26	nineteen	12731	2	nineteen
26	45	20	15200	2	ten
26	7	23	18143	five	7
26	48	20	13521	five	ten
26	eight	13	7625	6	15
26	eight	12	10,000	one	3
26	60	22	13550	3	13
26	2	25	14,000	6	2
26	54	25	11630	one	15
26	23	21	13740	five	nineteen
26	eleven	26	15273	7	13
26	eleven	18	12819	five	21
26	48	17	10792	one	sixteen
25	7	13	14429	6	eleven
25	7	20	18143	five	23
25	58	15	14121	five	eleven
25	32	17	11407	3	17
25	64	21	11,500	four	eleven
25	five	62	13,000	2	0
25	3	20	12,000	6	eight
25	49	21	14959	five	sixteen
25	59	18	11085	four	15
25	54	18	11667	one	ten
25	25	32	12773	one	18
24	eight	15	13750	one	22
24	25	34	14480	five	18
24	3	17	14334	7	0
24	51	21	12687	3	ten
24	74	24	13227	one	eleven
24	44	sixteen	10877	one	14
23	2	6	7,000	6	7
23	47	12	12469	four	14
23	five	18	10,800	2	3
23	34	17	10098	five	14
23	ten	31	14300	6	12
23	2	22	6000	2	five
23	3	thirty	10667	five	6
23	51	17	11059	five	12
23	thirty	20	11267	2	17
22	38	eleven	10079	2	9
22	eleven	nineteen	14455	7	nineteen
22	47	25	11915	3	sixteen
22	four	sixteen	10,500	four	23
22	13	14	9231	2	21
22	ten	26	12,400	one	21
22	82	18	12342	2	12
22	five	24	13,400	6	one
21	64	17	13250	four	12
21	6	14	11334	2	four
21	9	sixteen	7445	3	eight
21	four	17	11,500	one	five
21	sixteen	12	10625	2	eight
20	one	13	2000	7	four
20	2	eight	7500	3	five
20	67	18	10598	2	13
20	13	13	9924	7	18
20	7	nineteen	10858	2	2
20	45	13	10600	2	14
20	12	23	10167	four	0
nineteen	7	14	13143	6	17
nineteen	3	9	7,000	6	20
nineteen	ten	14	9200	four	21
nineteen	24	sixteen	9959	five	9
18	four	five	10,500	four	6
18	15	20	8267	3	20
18	3	6	6667	7	one
17	four	four	6000	four	one
17	2	23	10,000	four	four
17	29	13	6823	3	9
sixteen	12	ten	8,000	2	23
15	6	eight	8500	3	22
13	ten	eight	7100	7	20
13	five	26	5200	one	four
6	one	four	2000	one	6

Urgent news! The best results by rating are given by the publication on Tuesday at one in the morning! And in any case, do not publish on Monday at 4 or 6 am - your post will be doomed!

The class. Now let's see when it’s better to post to get more comments . I will not stick a sign, a lot of HTML code - I’ll just say that Tuesday morning is again the leader, and worst of all is Thursday morning and Wednesday six in the morning.
Viewed - better on Tuesday at one in the morning, worse on Monday at 6 in the morning.

What is this magic time for - Tuesday at night? Maybe those who are "in the subject" only at this time and published? Let's check:

 SELECT * FROM posts WHERE WEEKDAY(created)+1=2 AND HOUR(created)=1

The answer is already obvious to many:

Think before opening the spoiler

id	created	rate	comments	views
273403	2016-01-19 01:59:00	26	28	34,000
269577	2015-10-27 01:19:00	272	561	82000

From here we make logical conclusions on when it is best to publish an article. Sorry, another spoiler - for those who are not attentive and immediately decided to open the findings. I recommend to think again over the question before opening the spoiler.

Secrets of the universe, without SMS and registration

If you think a little more and pick up various options for sampling, it becomes clear that any “useful output” does not depend on the time of publication of the article. On any given day and time, an article may receive a number of ratings, views, and comments depending solely on its quality. My own hypothesis was confirmed.

It is best to publish an article - when you have a good, interesting and beautiful article ready. With love, your captain.

PS The author is well aware of the huge number of pogreh in this study, and it claims to be a joke rather than statistics. But in the joke there is a hint. It was not without reason that the Toaster, quite seriously, was asked a question and answered about the best publication time. Colleagues, write quality texts and do not worry about anything else.

PS2. Judging by the instantly arrived minuses, the revelers really thought that now they would be told the best time to publish an article. Which again shows that the question and this answer are very relevant.

In fact, there is still the question of how the rating of an article depends on its audience reach - but this is a fairly obvious topic, and does not apply to what we wanted to find out in this particular topic.

Thank you all, all the good articles! We meet here, at the very optimal time that we have identified.

UPD . A continuation and partial refutation of this article is in the second part .

Source: https://habr.com/ru/post/279607/

All Articles

HabroStatistics - at what time is it better to publish an article on Habré?

More articles: