Below you'll find for download various papers, slides, and other sorts of materials. They are in pdf files unless otherwise noted by a filename extension.
→ Note: New content is added (a)periodically. You may need to right click to download.
2018 ASA Conference on Statistical Practice Deep Learning Presentation Slides (February 2018, v2) a talk about deep neural networks that includes background on current technologies and a database marketing/CRM example application. A "try it at home" Jupyter Notebook with Tensorflow methods is available, below.
CSP 2018 DNN Binary Classifier Example v4.ipynb A Jupyter Notebook with a runnable and simple deep neural network classifier implemented in Tensorflow. "Try it at home" supplement to the talk "Scalable Deep Learning" at the ASA Conference on Statistical Practice, Portland OR USA, February 2018.
2017 Joint Statistical Meetings preliminary slides (8/1/2017 v7) for "Predicting Market Segment Membership Using Deep Learning," a talk about using deep neural networks that includes background on current technologies and a database marketing/CRM example application.
2016 Joint Statistical Meetings Proceedings paper about using "conditional random field" models to do part of speech tagging and named entity recognition with the text data in TripAdvisor reviews.
2016 Joint Statistical Meetings slides for "Room for Improvement: Aspect-Specific Statistical Opinion Mining of Online Hotel Reviews" presentation that goes with the above proceedings paper.
Bacon, L. & Lenk, P. (2012) "Augmenting discrete-choice data to identify common preference scales for inter-subject analyses", Quantitative Marketing and Economics, 10(4), 453-474.
2012 Joint Statistical Meetings Proceedings paper about specifying and applying models for behavioral engagment on a social media sharing platform and in a server cluster environment. (Key words: user engagement, Hierarchical Bayes, Hadoop, IRT, probit links, MCMC, CRM, marketing, big data, social media)
2012 Joint Statistical Meetings slides for "Capturing Heterogeneity in Engagement on a Social Media Platform" presentation that goes with the proceedings paper, above.
2012 working paper about augmenting Max-Diff and Choice-Based Conjoint models to enable inter-respondent comparisons. An early version of the article published in Quantitative Marketing and Economics , December 2012. (Contact us regarding the published article at .)
2010 Joint Statistical Meetings slides for a presentation about involvement (i.e. engagement) "trajectories" of users of a social media platform.
2010 Google Tech Talk slides for a presentation about using serious and purposive games to solve hard forecasting and machine learning problems, generating solutions to hard planning problems, new product development, and other such things.
2009 Joint Statistical Meetings Proceedings paper about user engagement and sales conversion on a social media platform by applying IRT-type models. (Key words: user engagement, IRT, Hierarchical Bayes, CRM, engagement)
2009 Sawtooth Software Conference Proceedings paper about game design and using games for marketing research. Included is a summary of an on-line linguistic discussion game for solving hard problems and generating new ideas.
2009 INFORMS Marketing Science presentation slides for a talk about using game design principles for marketing research. Includes empirical findings about the frequency of participation in discussion games as described by power laws.
2008 Marketing Research Magazine article about extending choice-based models (e.g. for Max-Diff or choice-based conjoint data) by incorporating additional response data so that results of different respondents can be compared directly.
2008 Joint Statistical Meetings Proceedings paper about B2B predictive CRM and and cross-selling based on likely response to offers.
2008 Marketing Letters article about modeling choices and preferences using large amounts of data. Based on a two-day workshop at the 2007 Choice Symposium held at Wharton.
2007 Sawtooth Software Proceedings paper about making Max-Diff and choice-based conjoint suitable for between respondent comparisons by augmenting choice models with additional response data.
2004 ART Forum slides for presentation about text analysis and natural language processing. Included is a summary of a IRT-like model that combines ratings data with processed verbatim comments to get to an improved customer satisfaction metric.
2002 Marketing Letters article about market structure and bundling based on a two-day workshop at the 2001 Berkeley Choice Symposium held at Asilomar, Pacific Grove CA in 2001.