{"id":1376,"date":"2017-07-13T13:40:32","date_gmt":"2017-07-13T13:40:32","guid":{"rendered":"http:\/\/kourentzes.com\/forecasting\/?p=1376"},"modified":"2017-07-13T13:47:11","modified_gmt":"2017-07-13T13:47:11","slug":"isf2017-presentation-call-centre-forecasting-using-temporal-aggregation","status":"publish","type":"post","link":"https:\/\/kourentzes.com\/forecasting\/2017\/07\/13\/isf2017-presentation-call-centre-forecasting-using-temporal-aggregation\/","title":{"rendered":"ISF2017 presentation: Call centre forecasting using temporal aggregation"},"content":{"rendered":"

This is joint work with Devon K. Barrow and Bahman Rostami-Tabar and is an initial exploration of the benefits of using Multiple Temporal Aggregation<\/a>, as implemented in MAPA<\/a> for call centre forecasting. The preliminary results are encouraging. More details in the attached presentation.<\/p>\n

Abstract<\/strong><\/p>\n

With thousands of call centres worldwide employing millions and serving billions of customers as a first point of\u00a0contact, accurate scheduling and capacity planning of resources is important. Forecasts are required as inputs for\u00a0such scheduling and planning in the short medium and long-term. Current approaches involve forecasting\u00a0weekly demand and subsequent disaggregation into half-hourly, hourly and daily time buckets as forecast are\u00a0required to support multiple decisions and plans. Once the weekly call volume forecasts are prepared, accounting\u00a0for any seasonal variations, they are broken down into high frequencies using appropriate proportions that\u00a0mainly capture the intra-week and intra-day seasonality. Although this ensures reconciled forecasts across all\u00a0levels, and therefore aligned decision making, it is potentially not optimal in terms of forecasting. On the other\u00a0hand, producing forecasts at the highest available frequency, and aggregating to lower frequencies, may also not\u00a0be ideal as very long lead-time forecasts may be required. A third option, which is more appropriate from a\u00a0forecasting standpoint, is to produce forecasts at different levels using appropriate models for each. Although\u00a0this has the potential to generate good forecasts, in terms of decision making the forecasts are not aligned, which\u00a0may cause organisational problems. Recently, Kourentzes et al. (2014)<\/a> proposed the Multiple Aggregation\u00a0Prediction Algorithm (MAPA), where forecasting with multiple temporal aggregation (MTA) levels allows both\u00a0accurate and reconciled forecasts. The main idea of MTA is to model a series at multiple aggregation levels\u00a0separately, taking advantage of the information that is highlighted at each level, and subsequently combine the\u00a0forecasts by using the implied temporal hierarchical structure. Athanasopoulos et al. (2017)<\/a> proposed a more\u00a0general MTA framework than MAPA, defining appropriate temporal hierarchies and reconciliation mechanisms,\u00a0and thus providing a MTA forecasting framework that is very flexible and model independent, while retaining all\u00a0the benefits of MAPA. Given the high frequency, multi-temporal nature of the forecast requirements and the\u00a0subsequent planning associated with call centre arrival forecasting, MTA becomes a natural, but yet unexplored\u00a0candidate for call centre forecasting. This work evaluates whether there are any benefits from temporal\u00a0aggregation both at the level of decision making as well as at the level of aggregation in terms of forecast\u00a0accuracy and operational efficiency. In doing so, various methods of disaggregation are considered when the\u00a0decision level and the forecasting level differ, including methods which results in reconciled and unreconciled\u00a0forecasts. The findings of this study will contribute to call centre management practice by proposing best\u00a0approaches for forecasting call centre data at the various decision levels taking into account accuracy and\u00a0operational efficiency, but will also contribute to research on the use of temporal hierarchies in the area \u00a0of high\u00a0frequency time series data.<\/p>\n

Presentation<\/a>.<\/p>\n

Related Posts<\/H3>