.
The following page is devoted to the forecast of the COVID-19 pandemic. It shows predicted cumulative number of COVID-19 related deaths by countries. The predictions are periodically updated, according to the data in this repository.

Only the countries that are significantly affected by the pandemic are included in the analysis.

• First, we present the Table (link below) with the one-day forecasts for deaths (two and three days forecast development is in progress) for each country. We also show 1 day back-testing prediction error for each country (as a percent of today’s recoveries).
• After this table, we show the 1 day back-testing Figure based on today’s data and parameters. In the upper part of the Figure, we show the actual number of deaths and the 1 day back-testing prediction of this number for each country. The lower part of the Figure, shows the back-testing error of each country’s prediction as a percent of deaths in this country. We calculate the mean and standard deviation for the error daily (see the tables’ heading).
• The above Figures are followed by the pages with the 2 columns of Figures:
  1. [left column: ] the next day prediction of cumulative number of deaths (rounded to the nearest integer), and the change from the current day cumulative number (two and three days forecast development is in progress).
  2. [right column: ] the long term trend in the rate comparison for the decline/growth of deaths of each country with the decline/growth of deaths of another country with some similar characteristics (in many cases, a neighboring country). Concavity of these graphs helps to estimate when there will be no more sick people. The pairing of the countries is re-evaluated daily. This helps to estimate the longer term trend in the decline/growth of cumulative deaths.
• Table with the next day forecast and last day prediction error.
• The last day prediction error:

• The next day forecast and comparison of the rates of change of deaths in country’s pairs (multiple pages).

Note: In some cases, due to estimation error, the number of cumulative cases declines slightly. In these cases, we can assume that the number of cases is not expected to change.