Why R? & McKinsey Webinar - Uplift modeling for marketing campaigns
* Mateusz Zawisza - Senior Analytics Advisor to McKinsey & Company
* Armin Reinert – Senior Data Scientist at McKinsey & Company
Topic: Uplift modelling for marketing campaigns
Do you know, who your sleeping dogs’, lost causes’ and sure buyers’ types of clients are? Only by knowing, who your clients are, you can truly maximize ROI on your marketing campaigns. You will need the profit-driven analytics to help you segment your clients into those segments. Unfortunately, classical Machine Learning (ML) cannot distinguish between those segments. As a result, traditional Machine Learning cannot fully maximize ROI, leaving money on the table. Since, ML models are deployed commonly in banking, telco, retail, etc., it creates the opportunity to improve on that. How? Through making the analytics more profit-driven and employing the most recent & state-of-the-art analytical methods from the intersection of machine learning, causal inference, uplift modelling and heterogeneous treatment effect estimation, e.g. Causal Forrest, Generalized Random Forrest, meta-learners, Qini curve, Area under Qini, see Athey, et al. (2019), Chen, et al. (2020), Taddy (2019).
In the presentation, we will demonstrate the example of marketing campaign and show, how we should properly calculate the true profit, i.e. uplift, from our analytical models. We'll present the segmentation of clients into sleeping dogs, sure buyers, lost causes, necessary for ROI maximization. Finally, we will show snippets of R & Python codes implementing these techniques.
Mateusz Zawisza - Senior Analytics Advisor to McKinsey & Company
Senior Analytics Advisor to McKinsey & Company with over 10 years of experience in designing & implementing advanced analytics solutions for retail, telecommunication, banking and public sector in Poland and abroad. He’s also coordinating McKinsey Analytics efforts aimed at recruitment, capability building and development programs. He holds Master degree in Quantitative Methods & Information Systems from Warsaw School of Economics, where he also worked for 9 years as Research & Teaching Assistant, published the book in predictive analytics as well as over 15 research papers about applications of quantitative methods. In the past he taught courses in Econometrics, Machine Learning, Optimization & Simulations, and currently is a lecturer on postgraduate studies of „Big data and data engineering” at Warsaw School of Economics.
Armin Reinert – Senior Data Scientist at McKinsey & Company
Data Scientists with over 3 years of experience in developing advanced analytics solutions related to marketing and sales topics (mainly pricing and CRM) in online and offline retail, and telecommunication industries. Graduate of Erasmus University Rotterdam with MSc degree in Econometrics and Management Science. He is particularly passionate about Bayesian methods and causality topics.