11-07, 15:20–16:00 (US/Eastern), Winter Garden
Learn how to use data to unlock consumer behavior insights with PyMC-Marketing. This open-source package bundles various pre-built marketing (e.g., media mix modeling) and customer analytics (e.g., customer lifetime value) models. The functionality provided by PyMC-Marketing massively simplifies the data scientist’s job, but to fully take advantage, one must understand the business problems each model solves, the type of data each model requires, and (a bit about) how these models work.
Data and analytics are at the heart of modern business operations. However, implementing advanced analytics can be a significant challenge. PyMC-Marketing is an open source Python package that seeks to simplify the process of implementing both marketing and customer analytics models. In this talk, we will discuss three significant challenges and how PyMC-Marketing helps to address them. First, what data sources can and should be included in analyses? Second, what analytics models should be used? The field is full of complex, math-heavy models and PyMC-Marketing implements several different families of analytics models. Third, it is critical to understand how to close the loop, using model-provided insights to guide business decisions. Along the way, we will discuss the sophisticated capabilities of PyMC-Marketing, including the simple specification of Bayesian priors, the incorporation of hierarchical structure and lift-tests, functions to validate model results, model diagnostics, etc.
This talk is intended for a variety of professionals, including data scientists and anyone working in marketing tech, operations, and business analytics who are seeking information about marketing and consumer analytics. The talk focuses on using PyMC-Marketing via brief, illustrative vignettes. Audience members interested in more technical material (i.e., implementation details and math) will be pointed to relevant resources. Audience members are not expected to have a background in modeling or math. Attendees will gain an understanding of both the basic concepts behind the models available in PyMC-Marketing as well as the package’s newest, most exciting features (including time-varying parameters, budget optimization, geo-hierarchical models, the incorporation of lift tests, model validation, and more).
Agenda:
Introduction to business and customer analytics (5 minutes)
Introduction to PyMC-Marketing (5 minutes)
Brief tour of Media Mix Models (MMMs) (5 minutes)
Brief tour of Customer Lifetime Value (CLV) models (5 minutes)
Translating model insights into to business decisions (5 minutes)
The future of PyMC-Marketing (5 minutes)
No previous knowledge expected
Dr. Christian Luhmann is Chief Operating Officer at PyMC Labs, a data science consultancy specializing in solving complex data science problems for businesses. In this role, Christian oversees the day-to-day global business of the firm and ensures that the organization has the necessary coordination, communication, and operating processes. Prior to assuming the role of COO, Christian acted as Project Manager, managing PyMC Labs' global client relationships, helping deliver innovative solutions to meet clients’ businesses' objectives, and driving business development efforts.
Christian began his career in academia, earning a BS degree in Computer Science and a PhD in Psychology. He spent 16 years as a professor, focusing on research in behavioral economics and machine learning and teaching statistics and data science. This experience instilled a strong desire to help others learn from their data.