ABSTRACT
In this article, a set of common statistical models, namely, linear, logarithmic, inverse, quadratic, cube, complex, power, exponential, and logistic model have been fitted to data representing the number of infections with Covid-19 virus in Iraq from the beginning of the disease until now by using the principle of fuzziness by forming a fuzzy information system (FIS) by generating values belonging to the set of infected numbers to produce a classical set that takes into account the inaccuracy (certainty) in data collection, then testing the significance of the models that were appropriate using the F-test and the probabilistic value sigma, and the comparison between these models using the coefficient of determination R-2 and MSE to reach the best model that represents the data of infection with the Covid-19 virus. Then estimate the best among those models and to calculate the estimated values for the number of infections with the virus. It was concluded that the use of the principle of fuzziness in the fitting of the models led to an increase in the accuracy of these models and the mean squares error (MSE) for all the models that have been fitted is reduced. We also note that the best model in representing the data of infections with the Covid-19 virus is the Power model, which recorded the lowest MSE among all the models, followed by the Logistic, Compound, Exponential models with the same strength of fit, with the same MSE at all alpha-cut coefficients (0.0, 0.1, 0.5, 0.8) and that the models Cubic, Quadratic, Linear, Logarithmic, Inverse are not suitable for data on the number of infections with Covid-19 virus, and we also note that the best model that achieved a fit for the data was at the alpha-cut = 0.8 (MSE=0.223) and that the value of the coefficient of the determination R-2 of the Power model decreases as the cut-off factor increases and this indicates the accuracy of the appropriate model. We also notice that increase in one unit of time led to increase infection with Covid-19 with 1.456.