Based on reported cases we fit and functions to make predictions into the near future. More info and explanation

Cases | Exponential prediction | Logistic prediction | |
---|---|---|---|

2020-03-27 | 50,871 | 49,302 | 50,640 |

2020-03-28 | 57,695 | 57,401 | 56,439 |

2020-03-29 | 62,095 | 66,832 | 62,133 |

2020-03-30 | 77,812 | 67,589 | |

2020-03-31 | 90,596 | 72,698 | |

2020-04-01 | 105,480 | 77,377 | |

2020-04-02 | 122,810 | 81,579 | |

2020-04-03 | 142,986 | 85,285 |

The blue bar describes the reported cases. They are used to fit two functions using statistical-mathematical methods so that the distance between the function and the data is minimal. The red curve is an exponential function, uncontrolled growth would follow this course. The purple curve is a logistic function ( S-shaped). Here, an initial part with exponential growth is followed by a decline, which is what was observed in China.

Nobody knows what the outcome of future cases will be. Even if a turning point were to be reached, the expected cases would continue to rise thereafter, only more slowly than with the exponential function. But every estimate is inaccurate and the number of data points is limited. Therefore, the results can only be interpreted using confidence intervals (box above). Only values outside of these shaded areas show a statistically significant deviation. In general, any function can be fitted to the data. Whether a function matches and follows the data can be assessed by a so-called quality of fit. In the following, a linear and a exponential function was fitted to example data.

You can see with the naked eye that the exponential function follows the data better. In other words, the exponential model is better suited to describe the data. This can also be expressed by the quality of fit (97 compared to 77 percent for the linear fit).

Data is updated automatically as reported by
Johns Hopkins University. Last update: 3/29/20.

Terms of use/Impress