FORECASTING MODELING OF THE INPUT LOAD ON THE GOVERNMENT CONTACT CENTRE

Анотація

The article conducted a model study of the load on the call centre of the state budget institution for its more efficient work. The relevance of this study is of particular importance in the context of the growing need for an uninterrupted process of communication between companies and customers with an increase in the cost of labour resources, the cost of their software and hardware. This problem can be solved by ensuring a balance of the necessary and sufficient amount of labour resources. With an increase in the efficiency of use of labour resources, their costs will decrease and the index of customer loyalty and satisfaction will grow. For this purpose, based on statistical data, forecast models are constructed. As a result of the predictive analysis of the input load data, the entire sample was reorganized in this way: the initial data were divided only by the days of the week, and the same data was divided by time for all days of the week. The reason for this separation was the impossibility of constructing an adequate forecast model for the entire sample due to the overlap of fluctuations in both time and days of the week, as well as the presence of non-stationarity caused, as the study showed, by a change in data depending on time and days of the week. However, such non-stationarity in the data did not allow building either an autoregressive model or a Winter’s model. Autoregressive – due to the non-removable non-stationarity by the difference method, the Winter’s model – due to the small trend. The imposition of oscillations did not allow them to be accurately simulate. A sample autocorrelation and a private sample autocorrelation function were constructed and the structure of time series was revealed for each of the obtained samples. Seasonal ARIMA models based on the Box-Jenkins approach were built, the adequacy of the developed models was studied and forecasts of the possible input load were made. The results of the forecasting carried out in this work were used to develop recommendations for improving the efficiency of use of labour resources at the enterprise under study.