Covid cases prediction has become an important tool in today's world. If we are able to predict covid cases accurately, then we will be prepared well in advance for the outcome. So, I have tried to predict covid cases for California state. The covid cases data can be found here.
I have used time series models like ARIMA, SARIMA and Holt Winters to predict covid cases and have used Thymeboost for outlier detection.
For ARIMA and SARIMA model, we require time series to be stationary and so, I have used ADF Test to check stationarity of time series.
In ARIMA model, we need to find the best Auto Regressive and Moving Average term. So, I have used ACF and PACF plots to find best Auto Regressive and Moving Average term.
A good time series model has uncorrelated residuals and to analyse residuals correlation, I have used Durbin Watson test.