ABSTRACT
This chapter investigates how a selection of multiple drivers jointly influences the KPI. The models considered in this chapter are explicitly suited to accommodating the simultaneous effects of multiple drivers. This is relevant for many data science projects because it allows for ranking the drivers of the KPI according to their impact so that decisions can prioritize the most influential drivers. For example, when allocating a media spending budget, it is relevant to know what channels elicit the strongest consumer response. This requires modeling consumer response as a function of the expenditures on each of these channels. Traditional, theory-driven models have a long tradition in marketing analytics. The steps for building a traditional theory-driven model are extensively described in Leeflang et al. They explain that building a model proceeds in four steps: specification, estimation, validation: checking assumptions, and use.
