STATISTISK ANALYS AV ETT CALLCENTER - MODELLERING AV INKOMMANDE SAMTAL

Sammanfattning: Statistical theory suggests that call arrivals to a telephone exchange follow a Poisson process and that time between successive calls is exponentially distributed. Previous studies indicate that the average number of arrivals changes over time, and thus follow a time-inhomogeneous Poisson distribution. The purpose of this thesis is to assess if the characteristics of daily calls to a call center operated by ISS Facility Services AB (ISS), is consistent with statistical theory and previous studies. The methods used include the Kolmogorov-Smirnov test, computer simulation, the Ljung–Box test, and linear regression. Our results indicate that daily calls appear to be a time-inhomogeneous Poisson process, but that that the time between successive calls is not always exponentially distributed. However, it is still concluded that call arrivals to ISS can be modeled as a single time-inhomogeneous Poisson process. According to our simulations, the way in which calls are transmitted between the main telephone exchange and the different call centers should skew the probability distribution to the right. Yet, the results of the simulations do not seem to be consistent with the empirical distribution. Moreover, an increased call volume might also distort the distribution.