TY - JOUR
T1 - Topological data analysis in digital marketing
AU - Lakshminarayan, Choudur
AU - Yin, Mingzhang
N1 - Publisher Copyright:
© 2020 John Wiley & Sons, Ltd.
PY - 2020/11/1
Y1 - 2020/11/1
N2 - The ubiquitous internet is a multipurpose platform for finding information, an avenue for social interaction, and a primary customer touch-point as a marketplace to conduct e-commerce. The digital footprints of browsers are a rich source of data to drive sales. We use clickstreams (clicks) to track the evolution of session-level customer browsing for modeling. We apply Markov chains (MC) to calculate probabilities of page-level transitions from which relevant topological features (persistence diagrams) are extracted to determine optimal points (URL pages) for marketing intervention. We use topological summaries (silhouettes, landscapes) to distinguish the buyers and nonbuyers to determine the likelihood of conversion of active user sessions. Separately, we model browsing patterns via Markov chain theory to predict users' propensity to buy within a session. Extensive analysis of data applied to a large commercial website demonstrates that the proposed approaches are useful predictors of user behavior and intent. Utilizing computational topology in digital marketing holds tremendous promise. We demonstrate the utility of topological data analysis combined with MC and present its merits and disadvantages.
AB - The ubiquitous internet is a multipurpose platform for finding information, an avenue for social interaction, and a primary customer touch-point as a marketplace to conduct e-commerce. The digital footprints of browsers are a rich source of data to drive sales. We use clickstreams (clicks) to track the evolution of session-level customer browsing for modeling. We apply Markov chains (MC) to calculate probabilities of page-level transitions from which relevant topological features (persistence diagrams) are extracted to determine optimal points (URL pages) for marketing intervention. We use topological summaries (silhouettes, landscapes) to distinguish the buyers and nonbuyers to determine the likelihood of conversion of active user sessions. Separately, we model browsing patterns via Markov chain theory to predict users' propensity to buy within a session. Extensive analysis of data applied to a large commercial website demonstrates that the proposed approaches are useful predictors of user behavior and intent. Utilizing computational topology in digital marketing holds tremendous promise. We demonstrate the utility of topological data analysis combined with MC and present its merits and disadvantages.
KW - Markov chains
KW - clickstreams
KW - computational topology
KW - landscapes
KW - persistence diagrams
KW - silhouettes
UR - http://www.scopus.com/inward/record.url?scp=85088590553&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85088590553&partnerID=8YFLogxK
U2 - 10.1002/asmb.2563
DO - 10.1002/asmb.2563
M3 - Article
AN - SCOPUS:85088590553
SN - 1524-1904
VL - 36
SP - 1014
EP - 1028
JO - Applied Stochastic Models in Business and Industry
JF - Applied Stochastic Models in Business and Industry
IS - 6
ER -