Leo González Pérez
07/24/2020 - 18:12
- Clarín.com
- Society
A week ago, on Friday, July 17, the Ministry of Health of the Nation registered 4,518 new covid-19 patients. The number was at the time the highest number of infected in a day. But, in addition, the number represented an increase of 894 cases in relation to the (3,624) reported the previous day. That daily jump was also record . On Saturday, July 18, the new cases were 3,305: the step, in this case descending (-1,213 cases), was also record. Always comparing one day with the previous one: on Sunday there was another upward shake, on Monday a decline and on Tuesday, July 21, the hitherto largest jump occurred from one day to the next in the total of new cases: 1,407. This "saw" - with a clear upward trend - caused the national government to begin reporting, in its morning video reports, the average of new cases per day for the last seven days , what the number experts know as the moving average (or "rolling average") .
Paying attention to this variable more than to the number of new daily cases, said the mathematicians consulted by Clarín , is an appropriate measure.
On the morning of this Thursday, Carla Vizzotti, Secretary of Access to Health of the Nation, and Alejandro Costa, Undersecretary of Health Strategies, reported that although the number of new cases of covid-19 in the country were that day 5,782, the average of new cases for the last seven days was 4,392. This Friday, they noted that the new patients were 6,127, but that the daily average for the last week was much lower: 4,749 .
Rodrigo Castro, professor at the UBA Faculty of Exact Sciences, Conicet researcher and director of the Laboratory for Simulation of Discrete Events, explains that the rolling average is commonly used to eliminate "noises" typical of some seasonality, which may be weekly, monthly, yearly, depending on what is being measured. "In the case of the covid-19," says Castro, "this seasonality can be delays in data recording during the weekends , which causes loads to accumulate towards the beginning of the following week. So, with a window seven-day moving average is calculated in that window and that should smooth the 'artificial' jumps in the variable being measured. "
Coronavirus in Argentina
Data source: Ministry of Health of the Nation Infographic: Clarín
Moving averages, the expert warns, present the risk that if in those seven days there was a "true" abrupt change, it could be smoothed out in the data record, which would not be desirable.
In the same sense as Castro, Pablo Groisman, a mathematician and researcher at the Faculty of Exact Sciences of the UBA and Conicet, explained that for cases such as that of the current covid-19 pandemic, if they average six or seven days, the Effects that "dirty" the curve with things that do not represent what you want to see are offset. "And what will remain will be a smooth curve that allows us to see the phenomenon that interests us, " he explained.
Guillermo Durán, director of the UBA Calculation Institute and adviser to the Government of the province of Buenos Aires in relation to the pandemic, agreed with his colleagues regarding the usefulness of moving averages. And he added, already in relation to the observed data, that the outlook is not very reassuring . "Neither the City of Buenos Aires nor the Conurbano are well, clearly," he said. And he added that he believes that the situation is complex, and not very different from what the people of numbers have been saying that could happen if the tracking of the links of the patients was not intensified and mobility was reduced.
For his part, Jorge Aliaga, PhD in Physics and former dean of the UBA Faculty of Exact Sciences, who has long applied moving averages to statistical information on the pandemic, warns that the national rolling average may hide the particularities of each region .
LGP