User talk:Malanoqa

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search
Welcome to Wikimedia Commons, Malanoqa!

-- Wikimedia Commons Welcome (talk) 15:50, 15 October 2013 (UTC)[reply]

Data

[edit]

Please correct the data in the table at File:West Africa Ebola 2014 12 Reported Cases per Week Total.png as needed. I tried to estimate the values from the chart, so many of the values could be slightly incorrect. See also the note about the 3792 additional cases that were added in the 29 October report. This also affects the Liberia chart, which also needs to be corrected. Liberia had a total of 867 new cases reported for the three week period ending on 25 October. Delphi234 (talk)

Hi Delphi234, I will check.--Malanoqa (talk) 08:13, 16 November 2014 (UTC)[reply]
Hi Delphi234, The bar graph as I supposed it to be, directly reflected the numbers in the table. My intention was not to make additionally changes other then interpolation. Updating the graphs is a lot of work, and I am happy you made a detailed look into this. And it is also good to know, that you can also update this bar graphs. Thanks for watching and supporting!--Malanoqa (talk) 08:21, 16 November 2014 (UTC)[reply]
Thank you for your help too! Delphi234 (talk) 20:35, 16 November 2014 (UTC)[reply]
Hi Delphi234, I now updated the bar graphs again. As the peak for Liberia is still in the table of the en:Ebola_virus_epidemic_in_West_Africa, I was not able to remove this, without using numbers I got from another source. But I added an explanation in the graphs that have this jump. Can you explain to me how you got the numbers that do not have the peak? If we use a different source for this numbers we have to discuss this on the talk page of the main epidemic article (en:Ebola_virus_epidemic_in_West_Africa), with kind regards, --Malanoqa (talk) 20:20, 24 November 2014 (UTC)[reply]
I used the WHO situation report. It gave the new cases over the previous 21 days, and I subtracted the data for the first two weeks of that period which left the new cases for the previous 7 days. For Liberia there were 867 new cases reported for the three week period ending on 25 October, so you can subtract the cases during the first two of those weeks to get the new cases for the third week. Delphi234 (talk) 04:07, 25 November 2014 (UTC)[reply]
Hi Delphi234, thank you for explaining. But we would be forced to correct the data presented in the tables first. In your calculation you reduced the value for this week, but you did not increase values in the previous weeks. I would leave therefore the weekly graphic as it is now with the additional remark that explains the peak. So OK? If not, please open a discussion on the talk page of en:Ebola_virus_epidemic_in_West_Africa. Thank you very much for you support, --Malanoqa (talk) 06:19, 25 November 2014 (UTC)[reply]
The only way to add in the data to the previous weeks is if WHO tells us where it goes, something they are very unlikely to do. The chart is fine with partial data. The fact that the recent data is more accurate than the old data is not important. There is only an error of 28% and there are suggestions that the data is off by 100 or 200%. It would be way too artificial and way too prone to introduce errors to try to guess when the extra cases occurred. Delphi234 (talk) 04:49, 26 November 2014 (UTC)[reply]
Not really. The chart takes precedence over the table. While some charts base their data on material in an article that is not the best way to do it, the best way is to independently base the chart on reliable sources. It is not a good idea to perpetuate a fallacy in a chart when you know that the anomaly is not real, and is only caused by the WHO going back and adding in hundreds of cases from earlier, and even saying in the report that they did not occur during the last seven days. Can you list the specific data that was used to create the Liberia chart, like was done here? Delphi234 (talk) 21:02, 25 November 2014 (UTC)[reply]
Hi Delphi234, you replaced only the data for the week ending 2 November with a different value. But if we base the graph on other data, we need a consistent set of data. I will open the discussion on en:Ebola_virus_epidemic_in_West_Africa now. With kind regards, --Malanoqa (talk) 22:42, 25 November 2014 (UTC)[reply]
It is my understanding that there is only one source for the data - the WHO situation reports. It is pointless to use data from the article if it is clearly wrong. I apologize for removing the last week of data from the chart, but just use the data that is in the table, and correct the table if necessary. A week of missing data was better than a chart that showed wildly erroneous data (and pointless to add a note saying this data point is meaningless, when the correct data is available - charts often have spurious anomalies, but this one does not need to be there). But by all means correct the data in the article as well, but unless it is challenged or controversial there is no need for any talk page discussion. There is nothing controversial about the data, other than the fact that there could be more cases, more deaths, than have been identified. I have noticed that the number of cases and deaths has occasionally been reduced, presumably because possible cases/deaths have been removed when they were found to not be ebola related. Delphi234 (talk) 23:04, 25 November 2014 (UTC)[reply]
Hi Delphi234, there is a need to discuss this on the talk page of the article. The reason is, that we both do not agree on how to handle this. And I doubt that we will find a solution here. Please explain your arguments now on the talk page Handling of "jump" in numbers for Liberia in week 26 October, as my personal talk page on Wikimedia is not the best page to solve this. With kind regards, --Malanoqa (talk) 23:24, 25 November 2014 (UTC)[reply]
I am not sure why there would be any disagreement. The number of cases over the last three weeks in west Africa in Table 1 was 2966, 666 in Guinea, 867 in Liberia, and 1433 in Liberia. Since the previous report there were 3562 additional cases, and since the report three weeks earlier (Oct. 8) there were 5665 new cases, but the Oct. 29 report says that 3792 of those took place during the entire epidemic and not during the most recent period. By the way, if you do the math, Table 1 on Oct. 29 has one column mislabeled, it should say "Cases in past 21 days/total cases". The numbers are too big for the 21 days column to be the one that is mislabeled. We can not go back and distribute the 3792 additional cases into the correct weeks without the WHO doing that, and it is highly unlikely they will have any reason to do so. So all we can do is report the number of new cases reported as happening each week, and we certainly can not include the 3792 cases that occurred earlier as if they had occurred the week ending October 26 when we know that that is not when they did occur. We can include the jump in cases in the chart of total cases, as if they occurred that week, but not in the chart showing weekly new cases. They are also not included in the File:2014 Ebola Outbreak Monthly Cases per day.svg, but were included until I saw the note about them occurring earlier, and the chart was corrected. Delphi234 (talk) 00:19, 26 November 2014 (UTC)[reply]
Guinea         Total   21 days Report  21 days column divided by total column
Confirmed	1391	332	24%	24%
Probable	199	19	10%	10%
Suspected	316	315	100%	100%
All		1906	666	35%	35%
Liberia				
Confirmed	2515	35	1%	1%
Probable	1540	363	24%	24%
Suspected	2480	469	19%	19%
All		6535	867	13%	13%
Sierra Leone				
Confirmed	3700	1187	32%	32%
Probable	322	14	4%	4%
Suspected	1213	232	19%	19%
All		5235	1433	27%	27%
Total		13676	2966	22%	22%
It looks like you got your answer. No one wants the graph. But actually there is nothing wrong with it if you just remove the extra data from earlier weeks that is included in the week ending around Oct. 26. There has been quite a bit of discussion of plotting cases/deaths per day, and unless it is averaged for a week or a month it jumps all over the place. It is correct that I can not reply there, but it is totally false that I am either a puppet or a sock. This is my main account, and I only act in a manner that is permitted. We allow alternative accounts because they are necessary. Delphi234 (talk) 04:12, 26 November 2014 (UTC)[reply]
By the way, even if the chart is removed from the article, I strongly encourage you to continue to update it as it is a valuable chart. Creating charts is mostly a thankless task but when you get feedback like that you know that your work is appreciated. My only suggestion is that it would be better as an SVG, which I can create if you can supply the data. See for example the relatively similar File:US Solar Generation-2013.svg I started to work on weekly cases/deaths but do not have a complete dataset, and switched to doing it monthly. SVG files are just plain text and are easy to update with just a text editor. The only tricky part is when the scale needs to be changed. Delphi234 (talk) 04:41, 26 November 2014 (UTC)[reply]
Hi Delphi234, no there is a disagreement. When you follow our discussion, you see that we both just repeat our arguments. We do not come close together to approach a common opinion. So we need other opinions for this. Sadly the editors in the talk page, did not ask the main question. They said remove, which circumvent the problem, but does not solve our conflict. It would surely be better to use the weekly graphics from WHO, but for this they have to be under Wikimedia Commons Licence. And I have no time to organize this. And if you ask the editors whether a graphic should stay or be removed you read normally very different opinions in this artice. So three remove at the start, does not surprise me. With kind regards, --Malanoqa (talk) 18:47, 26 November 2014 (UTC)[reply]

What makes the chart so valuable is that the number of new cases per day and deaths per day provides an accurate sense of whether the outbreak is increasing or decreasing. To do that we need to plot only the new cases that occurred each week, and the deaths that occurred each week. If on 1 Nov. the WHO reports that there were 7,000 previously unreported deaths that occurred during 2014, we can not add that to the chart as saying that there were 7,000 deaths that occurred on 1 Nov, because that is not when they occurred. The only time we can include any data is when we do know when they occurred. The fact that we now know that the data prior to the 29 Oct report should have included 28% more cases, but not any more deaths, is simply not anything we can do anything about - we already have heard that the number of cases could be 2 or even 3 times higher than reported. All we can do is report what we know, and the chart does show the growth and decline in the outbreak. Or will when it does finally end. If you make the chart useful and maintain it I promise you that everyone will want it in the article. But when it includes errors not so likely. There are also currently 46 different language wikis that have articles about the outbreak that could use the chart. Delphi234 (talk) 20:38, 26 November 2014 (UTC)[reply]

Hi Delphi234, yes a better graph has more chances to stay in the article. And we also know how problematic the graph and the numbers are. People may tend to believe the graph they see, and they may not assess correctly how problematic the data is. Ebola is hard to diagnose and there are also reports that patients are hidden. Because people fear that patients that have for instance only Malaria are placed together with Ebola patients in a treatment center. To become infected. So I am always doubting whether graphs are good or bad. They may cause people to act, but this action might not be correct. With kind regards, --Malanoqa (talk) 21:05, 26 November 2014 (UTC)[reply]
Charts, and an encylopedia, for that matter, are simply a window upon the world, and simply helps people to know factual information. It does not necessarily prompt any action. Charts are particularly valuable because a picture is worth a thousand words and helps to visualize the data and see trends. Can you provide the actual data used? Reading your 24 Nov. update I get different numbers, than are on the chart page, but it is pretty inaccurate reading a chart. Or you can just update the table on the chart page. The number that is most off is for 2 Nov. Thanks. Delphi234 (talk) 21:56, 26 November 2014 (UTC)[reply]
Date estimated
from
chart
table actual
07/13/14 130 126
07/20/14 116 121
07/27/14 229 228
08/03/14 349 345
08/10/14 233 238
08/17/14 442 443
08/24/14 622 625
08/31/14 727 727
09/07/14 655 657
09/14/14 969 974
09/21/14 919 923
09/28/14 925 927
10/05/14 837 839
10/12/14 965 965
10/19/14 936 937
10/26/14 N/A 1065
11/02/14 815 480
11/09/14 1053 1056
11/16/14 1046 1047
Hi Delphi234. I must really apologize, that I do not check your numbers and calculations in detail. But I am propably not the only person active in the current epidemic that is flooded with topics. I have simply no time to develop a new consistent number base for the graphs. It is already time consuming to update 9 to 13 graphs every few days. Do not forget, many of us are volunteers, that have work and family to care for too. And to collect and check all this numbers is also a big task. Now starting to collect consistent different numbers for certain graphs would be even more time consuming. And I have other topics during this epidemic, that I rate to be more important. Some chapters need more reliable and helpful informations. And as people are just in this moment developing the procedures and technologies for this, we are not even able to write there something better that is also well based on reliable references. Something that is really helpful. And to me this peak is also a good sign to the visitor, of how problematic all this numbers are. So why removing it, when I can also add a checkbox that explains it?--Malanoqa (talk) 22:25, 26 November 2014 (UTC)[reply]
It is my turn to apologize. I was misinterpreting this chart as being "new cases per week", but it is labeled "reported cases per week", so the large bar can remain, but is useless in determining "new cases per week" which would be an indicator of the rise and fall of the outbreak, and since that week is completely off scale hides the real data on new cases per week. So the chart is just a report on the WHO reporting and does not actually indicate the progress of the disease, and is not particularly of any interest, at least not to me. Please carry on, though. In Table 1.[1] there were 1828 new cases over the 21 days prior to 2 Nov. in Guinea, Liberia, and Sierra Leone, and none that I can see anywhere elsewhere, so the numbers for 19 Oct, 26 Oct, and 2 Nov should total 1828, and if you look at Table 1.[2] the total for the weeks ending 26 Oct, 19 Oct, and 12 Oct should be 2966. The 12 Nov report, covering the period up to 9 Nov, gives in Table 1.[3] 2002 cases during the weeks ending 26 Oct, 2 Nov, and 9 Nov. These numbers are complicated by suspected cases either becoming confirmed cases or being removed as not being ebola. All I was saying in providing the data is you must have used some sort of spreadsheet to plot the data, and I just wanted to see the column of data that you used, so that it could be used in other charts. Delphi234 (talk) 22:38, 26 November 2014 (UTC)[reply]
Good news though. The WHO has just released the patient database for new probable and confirmed cases, going all the way back to January. Click on "View data". Delphi234 (talk) 01:51, 27 November 2014 (UTC)[reply]

German Chart COVID-19

[edit]

Thank you for the chart. Please go on. And keep care. -- Willi The Kid (talk) 13:51, 21 March 2020 (UTC)[reply]

Hi Willi The Kid, I updated the charts. You may also be interested in https://github.com/RainerWinkler/COVID-19-plot, it contains a lot of plots and the option to generate own plots with Python. I wish you to stay healthy, Malanoqa (talk) 17:44, 21 March 2020 (UTC)[reply]
Thank you :-) Willi The Kid (talk) 10:25, 18 April 2020 (UTC)[reply]

Two “Gesamt” curves

[edit]

Questions about File:Coronavirus Germany cumulated logarithmic.png: Obviously, they can't be both “Gesamt” (total) – is the gray curve the real total number of cases reported to the RKI, and the black one just a subset? (Antwort gerne auch auf Deutsch.) ◀ SebastianHelm (talk) 12:10, 6 April 2020 (UTC)[reply]

Hi SebastianHelm, The Robert-Koch-Institut changed from manual to electronic reporting. For a few days the reported two values. This is reflected in two "Gesamt" values. BRMalanoqa (talk) 13:10, 6 April 2020 (UTC)[reply]
Thank you. BTW, it seems the first two common days (10.3. and 11.3.) they had the same values, since the two curves overlap. Is that right? https://www.rki.de/DE/Content/InfAZ/N/Neuartiges_Coronavirus/Situationsberichte/2020-03-10-de.pdf?__blob=publicationFile says “davon wurden bisher 654 elektronisch an das RKI übermittelt und hier validiert”. The next day, too, the electronically reported cases seem to be much less than the manual ones. ◀ SebastianHelm (talk) 17:25, 12 April 2020 (UTC)[reply]