Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume

Detalhes bibliográficos
Autor(a) principal: Peixoto, Ricardo de Carvalho Lourenço Beirão
Data de Publicação: 2023
Tipo de documento: Dissertação
Idioma: eng
Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
Texto Completo: http://hdl.handle.net/10362/151751
Resumo: Project Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
id RCAP_a94977cde8fd15d70eb85c94a3a1479a
oai_identifier_str oai:run.unl.pt:10362/151751
network_acronym_str RCAP
network_name_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository_id_str 7160
spelling Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic VolumeMarketing mix modelingBayesian InferenceMulti-Touch AttributionProject Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data ScienceThe marketing budget, in the company, is allocated to strategically drive specific KPIs. The pandemic forced a shift in the investment strategy due to already optimal KPI results or restrictions in the capacity to further improve them. The company's growth is now expected to come, although not solely, from an increase in Traffic, a KPI with few to no experimental exercises. Specifically to the department, the focus was to understand the contribution of marketing to website traffic, its drivers and levers, and to make forecasting exercises viable. The marketing effectiveness problem is long-known, with no consensus on the methodology to apply. However, the need to include offline marketing channels and the increasingly strict privacy policies that could threaten the solution's longevity led to a Marketing mix model being developed to answer the stakeholder's needs. The availability of prior information, in the shape of Multi-Touch attribution insights, and the benefit of quantifying uncertainty for both parameter estimations and model estimates made a Bayesian approach to Marketing mix model the preferred choice, fitting the model equation with the NUTS algorithm available on the PyMC3 python library. Different prior specifications were tested, but the combination of a t-student for the likelihood to account for outlier effect in estimation, an exponential distribution for the error term, and informative priors for the adstock parameters were selected as the model with the most potential since it produced slightly worse NRMSE values but significantly better results for the MAPE metric. The sample size was increased to achieve convergence and better parameter space exploration, which resulted in an average Gelman-Rubin 1.02 for the parameters, narrower credibility intervals, lower average standard deviation values and a shift of around 12% in the posterior mean estimates for the parameters. The model performance for the predictive posterior distribution, replicas using parameter point estimates and out-of-sample predictions using the same point estimates are, 0.046 NRMSE and 0.049 MAPE, 0.046 NRMSE and 0.047 MAPE, and 0.070 NRMSE and 0.118 MAPE, respectively. For the company, the model initiated the ongoing validation and finetuning process natural of MMMs providing the first independent view on website traffic and marketing effects while contributing to the scarce research on MMMs in the Sporting Goods industry and covid impact modelling variables for the same models.Jesus, Frederico Miguel Campos Cruz Ribeiro deRUNPeixoto, Ricardo de Carvalho Lourenço Beirão2023-04-12T14:24:52Z2023-04-102023-04-10T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10362/151751TID:203268024enginfo:eu-repo/semantics/embargoedAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2024-03-11T05:34:08Zoai:run.unl.pt:10362/151751Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-20T03:54:41.111422Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse
dc.title.none.fl_str_mv Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
title Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
spellingShingle Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
Peixoto, Ricardo de Carvalho Lourenço Beirão
Marketing mix modeling
Bayesian Inference
Multi-Touch Attribution
title_short Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
title_full Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
title_fullStr Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
title_full_unstemmed Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
title_sort Marketing Mix Modeling a Bayesian Approach to Assess Advertising Effectiveness on Website Traffic Volume
author Peixoto, Ricardo de Carvalho Lourenço Beirão
author_facet Peixoto, Ricardo de Carvalho Lourenço Beirão
author_role author
dc.contributor.none.fl_str_mv Jesus, Frederico Miguel Campos Cruz Ribeiro de
RUN
dc.contributor.author.fl_str_mv Peixoto, Ricardo de Carvalho Lourenço Beirão
dc.subject.por.fl_str_mv Marketing mix modeling
Bayesian Inference
Multi-Touch Attribution
topic Marketing mix modeling
Bayesian Inference
Multi-Touch Attribution
description Project Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
publishDate 2023
dc.date.none.fl_str_mv 2023-04-12T14:24:52Z
2023-04-10
2023-04-10T00:00:00Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/masterThesis
format masterThesis
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10362/151751
TID:203268024
url http://hdl.handle.net/10362/151751
identifier_str_mv TID:203268024
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/embargoedAccess
eu_rights_str_mv embargoedAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron:RCAAP
instname_str Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron_str RCAAP
institution RCAAP
reponame_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
collection Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository.name.fl_str_mv Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
repository.mail.fl_str_mv
_version_ 1799138135330783232

Registros relacionados