Exploring the use of machine learning and explainability in Marketing Mix Modeling

[EN] Marketing Mix Modeling (MMM) employs statistical techniques, typically linear regressions, to assess the impact of advertising expenditure on sales. Despite advancements in statistics and machine learning, the field of MMM has remained relatively unchanging due to a few reasons: (1) its primary...

Descripción completa

Detalles Bibliográficos
Autores: Kisilevich, Slava, Herrmann, Markus
Tipo de recurso: capítulo de libro
Fecha de publicación:2023
País:España
Institución:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:inglés
OAI Identifier:oai:riunet.upv.es:10251/201700
Acceso en línea:https://riunet.upv.es/handle/10251/201700
Access Level:acceso abierto
Palabra clave:MMM
Marketing Mix Modeling
Machine Learning
Explainability
SHAP
Descripción
Sumario:[EN] Marketing Mix Modeling (MMM) employs statistical techniques, typically linear regressions, to assess the impact of advertising expenditure on sales. Despite advancements in statistics and machine learning, the field of MMM has remained relatively unchanging due to a few reasons: (1) its primary focus on practical business applications, (2) the proprietary nature of MMM solutions by specialized companies, and (3) the difficulty in interpreting complex models beyond linear regressions for business purposes. Recently, there has been increased emphasis on the interpretability of complex machine learning models. To address this, model explainers such as SHAP have been introduced, enabling the application of non-linear machine learning algorithms in the realm of MMM. This provides a solution to the various issues associated with traditional MMM methods, including variable interactions, non-linear relationships, and interpretability. This presentation outlines a method for incorporating machine learning algorithms with explainability techniques in the context of MMM in the retail industry.