When considering different allocations of the marketing budget of a firm, some predictions, that correspond to scenarios similar to others observed in the past, can be made with more confidence than others, that correspond to more innovative strategies. Selecting a few relevant features of the predicted probability distribution leads to a multi-objective optimization problem, and the Pareto front contains the most interesting media plans. Using expected return and standard deviation we get the familiar two moment decision model, but other problem specific additional objectives can be incorporated. Factorization Machines, initially introduced for recommendation systems, but later used also for regression, are a good choice for incorporating interaction terms into the model, since they can effectively exploit the sparse nature of typical datasets found in econometrics.
[1]
Jasbir S. Arora,et al.
Survey of multi-objective optimization methods for engineering
,
2004
.
[2]
David B. Dunson,et al.
Bayesian data analysis, third edition
,
2013
.
[3]
Christoph Freudenthaler,et al.
Bayesian Factorization Machines
,
2011
.
[4]
V. Gallego,et al.
Sales forecasting and risk management under uncertainty in the media industry
,
2018
.
[5]
Steffen Rendle,et al.
Factorization Machines
,
2010,
2010 IEEE International Conference on Data Mining.
[6]
Chih-Jen Lin,et al.
Field-aware Factorization Machines for CTR Prediction
,
2016,
RecSys.