Predictive Modeling for Resource and Demand Management

Predictive modeling for resource and demand management is defined and the core underlying concepts, basic assumptions, and usual limitations are identified. Connections to decision-making contexts and measurable outcomes are established.

In predictive modeling, a target variable of interest is predicted from the values of a number of other variables. Choice of target variable and prediction destination shape the predictive modeling framework. For example, predicting the `mean number of customers' at `time t + one hour in the next day in an automated small food store' is less useful than predicting the number of customers in the restaurant. A small food store selling culinary take-away products, soft drinks and beer requires staff and shelf management to satisfy demand and not risking loss of sales or profit reduction due to excess waiting time. Demand prediction at exhaust shelf filling level avoids excess storage and preparing of meals without sales. Demand prediction could be a predictive model in resource management capable of proactively reverting resource demand fluctuation; or, a network theory predicting arrival rate to design resilient networks, planning services or studying epidemic spreading speed.