Covid-19

This is an automatically generated report using the latest information and does not validate any information before publishing the results. Only “signal” or at most “provisional” estimates will be provided.

It uses daily data provided by 2019 Novel Coronavirus COVID-19 (2019-nCoV) Data Repository by Johns Hopkins CSSE and makes no adjustments. Then, live data from Worldometers is used to update only the totals and see last hour data.

As forecast models have already appeared on this topic, we will try to make a dynamic model, which will be updated daily (so the current posting changes daily) starting from the modeling of better developed countries, that could apply to Romania’s forecasts. A dynamic model of confirmed cases should give us a better insight of the economic impact of the virus.

New information as at 2020-09-22 based on 31,517,087 confirmed cases worldwide (982,824 deaths, 651 deaths in the last 24h)

               Country     Sum LastDay TwoDays  Newly Deaths Recovered     Tests
 1:                 US 6896218   39334   52070 243335 206593   4398907 100582090
 2:              India 5562663       0   75083 167521  91173   4671850  66279462
 3:             Brazil 4591364   33324   13411  36416 139065   3992886  17900000
 4:             Russia 1111157    6109    6090  11084  19799    923699  43600000
 5:           Colombia  777537    7102    5359   6731  24746    662277   3499136
 6:               Peru  768895       0    6030  13800  31870    636489   3751583
 7:             Mexico  705263    4683    2917   4786  74949    510237   1620878
 8:              Spain  682267   10799   31428  11289  31034         0  11820505
 9:       South Africa  663282    1346     725   1906  16206    594229   4083757
10:          Argentina  652174   12027    8782  12625  14376    525486   1815738
11:             France  507150   10299   29237      0  31459     93538  10000000
12:              Chile  448523    1055    1194   1380  12345    425165   3084117
13:               Iran  429193    3712    3341   3605  24840    365846   3828330
14:     United Kingdom  406058    4936    4378   3671  41862         0  22616796
15:         Bangladesh  352178    1557    1703   1666   5044    262953   1862637
16:       Saudi Arabia  330798     552     492    561   4569    313786   6190822
17:               Iraq  327580    4724    3821   5055   8754    264988   2121304
18:           Pakistan  307418     532     582      0   6432    293916   3264216
19:             Turkey  306302    1692    1743   1767   7711    270723   9608370
20:              Italy  300897    1391    1350   1640  35758    220665  10679675
21:        Philippines  291789    1599    3447   2802   5091    231373   3462679
22:            Germany  277412    1852    1595   1793   9508    249500  14557136
23:          Indonesia  252923    4071    4176   4465   9977    187958   3032250
24:             Israel  193374    2445    3027  11316   1325    144963   3195429
25:            Ukraine  185890    2990    2781      0   3705     81670   2079482
26:             Canada  148855    1272    1833      0   9243    127788   6833661
27:            Bolivia  131453     467     310    537   7731     91556    292016
28:            Ecuador  127643     932     292   2249  11171    102852    399096
29:              Qatar  123917     313     228    258    212    121006    739091
30:            Romania  114648    1059     808   1767   4550     93558   2278125
31: Dominican Republic  109269     486     494    468   2074     83434    455525
32:         Kazakhstan  107450      76      67      0   1699    102064   2733455
33:             Panama  107284     474     607    706   2291     84437    447769
34:            Morocco  105346    2227    1376   2397   1918     88244   2457373
35:            Belgium  105226    1834    1097      0   9955     19039   2942559
               Country     Sum LastDay TwoDays  Newly Deaths Recovered     Tests

Note. Last four columns are from Worldometers. Newly represents the difference from live data to last day data.

Choosing a model for different country based solely on that country’s data.

The first step is to find the best ARIMA model for every country. Keep in mind that the confirmed cases depend on the tests done prior getting the results, not on the direct impact the virus develops naturally.

Note that Romania can be very small at the bottom due to development similar to US, but back in 2020-09-23

Romania’s case based on different models

Now we will try to forecast Romania’s evolution on different assumptions. The most important information is to test if the confidence intervals (dotted lines) get broken by the new results. The plot uses semi-transparent colors to allow for overlapping.

On 2020-03-25 Spain and Italy were poised to the same model (Italy+). US will surpass all other countries. Clearly all aver the world people are tested more than in the past. Romania’s cases will double before 2020-04-04.

On 2020-03-27 Germany and US were poised to the same model (US+). Romania has none of the models (neither US+ nor Brazil+).

On 2020-03-28 Russia entered in the picture, just behind Romania. I’ll bet they will have an unique model, don’t you agree?

On 2020-04-03 Romania started to test at normal capacity. The model should converge to a stable one.

On 2020-06-03 The analysis takes into consideration last 100 days.

As we reached 10 milion cases worldwide, Romania’s model got a drift!

On 2020-09-08 Italy was replaced by Brazil and Spain by India.

Updated: 2020-09-24 05:00:17

Valentin Cornaciu
Valentin Cornaciu
Actuary

My research interests include insurance, finance, audit, pricing.