Predictions on number of world-wide corona-virus cases over time

Raw data since: 2020-01-22, predicting since 2020-01-22, last sample:2020-07-13 07:30:22 n=13041698

Latest results. Purple is data, dots are predictions

The error reported in the figure (Err=...) is the rms error of a predictive model on the training
data for the model. The training data are the known daily (Worldometer) counts of
corona-virus cases, fat purple curve, since 22/1/2020.


  • Febr. 2020 Movie (.mp4) of global Covid-19 predictions (since 5/2/2020)

  • Actual and Modeled (tentative) New cases (may be too optimistic, is rarely pessimistic)

  • Overview over several countries (daily new cases)

  • Prediction errors (Ncases) best logistic model of a day as determined that day

    At the end of each day, the best model is selected for making a prediction for the next day.
    The actual count is determined at the next day from Worldometer corona virus report.
    No other data points are collected from reporting sites during the day, when computing the prediction.

    Additional tables

  • List of .csv Downloads (Global)

  • Table of prediction errors, Global

  • Modeling stability (coefficient of variation for rmse)

  • Explanation of the computational approach
    Lambert Schomaker - AI Dept. - Bernoulli Institute - University of Groningen