| dc.description.abstract | In this report we examine a data set from TrønderTaxi, which contains information about all phone calls their call center received from 1 March 2014 to 31 January 2016. We model the number of phone calls received per 30 minute time interval using two different models. The first model is a seasonal autoregressive integrated moving average model, or SARIMA model. The second model is a naive model, which treats all of the weeks of the data set as independent and identically distributed. We use both the SARIMA and naive models to make predictions of both one and two weeks into the future. The results show that the SARIMA model appears to make slightly better predictions for one week into the future, while the naive model is better for more distant predictions. | |
