In this thesis we described what a time series is, we gave theoretical background and established the tools and basis for its analysis. Also, we gave concept of the stationary process, ARMA(p; q) process, autorcovariance function, autocorrelation function (acf) and partial autocorrelation function (pacf) which have a crucial role and are a mighty tool in their analysis. We used elaborated theory on concrete and real data provided from anonymous source from Norwegian hospital using static program functions in R. The goal of the analysis was to model data using the ARMA process and to establish predictions according to it. We used differencing method on the starting data to remove its seasonality. Autocorrelation and partial autocorrelation functions provided us with order of the ARMA(p; q) process. Choosing model and using it on one part of the differenced data (we took the number of examinations in the first 35 time points), we made predictions on the modeled data for the next 4 days. We made a graphic preview of the differenced series and added the prediction on top of it. We can deduct that the ARMA process outputs relatively accurate prediction results. The predicted results will always provide only an approximation of the values, so they should not be taken as precise values.