Webinar July 20: Using Machine Learning and Data Analysis to Improve Customer Acquisition and Marketing in Residential Solar
July 3, 2017
Join a free webinar July 20 to learn how data analytics, machine learning, and social science can improve marketing and customer acquisition processes for residential solar.
High customer acquisition costs remain a persistent challenge in the U.S. residential solar industry. Effective customer acquisition in the residential solar market involves multiple moving parts: increasing referral rates, having a good ground game, and designing effective marketing campaigns.
A free webinar on Thursday, July 20, 2017, from 1–2 p.m. ET, will share exciting new research by the National Renewable Energy Laboratory (NREL), Sandia National Laboratories, and Vanderbilt University on how data analytics, machine learning, and social science can improve marketing and customer acquisition processes.
The webinar will feature the following three presentations:
Solar Technology Diffusion through Data-Driven Behavior Modeling
Kiran Lakkaraju, senior member of the technical Staff at Sandia National Laboratories
Increasing use of clean, renewable energy can help reduce dependence on fossil fuels. Given that residential energy use is 21% of energy consumption in the United States, there has been increased interest in understanding and predicting residential consumer behavior towards purchasing solar photovoltaic (PV) panels. Better prediction of consumer behavior can reduce customer acquisition "soft costs," which will reduce solar PV prices. This talk will provide an overview of efforts to develop a computational model of residential consumer behavior. Many factors influence this complex decision, from economic, attitudinal, demographic and technical. We have created an agent-based model that incorporates a wide array of these features to predict solar PV purchasing trends based on household level data from San Diego County. Predictive models can help inform decisions by allowing “what if” analyses. We will discuss how to use the predictive model to explore financial incentive schemes, and to study how different marketing approaches can increase adoption.
Yevgeniy Vorobeychik, assistant professor of Computer Science, Computer Engineering, and Biomedical Informatics at Vanderbilt University
Agent-based modeling has been extensively used to study innovation diffusion. We develop a novel methodology for data-driven agent-based modeling that enables both calibration and rigorous validation at multiple scales. We then use this model as a core piece of an algorithmic marketing framework aimed at promotion of an innovation (in our case, rooftop solar in the residential sector), considering problems such as door-to-door marketing and multi-channel marketing.
Using Social Science to Design Better Online Solar Advertisements
Michael Rossol, System Modeling & Geospatial Data Science Group in the Strategic Energy Analysis Center at NREL
Online lead generation for residential solar through email or web banner advertisements offers an intriguing opportunity to reduce customer acquisition costs because these methods have low variable costs and provide opportunities for targeted messages. However, two drawbacks of online marketing are low response rates and difficulty in understanding why any specific advertisement was successful. We demonstrate the use of short customer surveys that mimic how a customer might respond to typical online advertisements. These methods allow us to understand how and why different marketing approaches are effective. These types of experiments might provide a low-risk opportunity to evaluate various competing advertisement types, say, before launching a major advertising campaign.
Kiran Lakkaraju is a senior member of the technical staff at Sandia National Laboratories in New Mexico in the Cognitive Science and Applications group. He has more than 10 years of experience developing models of human behavior and social systems and has background in artificial intelligence, multi-agent systems, and computational social science. He holds a M.S. and Ph.D. in computer science from the University of Illinois at Urbana-Champaign. Lakkaraju’s research has been marked by extensive interdisciplinary efforts that bring together the social and computational sciences. His Ph.D. work was focused on creation and propagation of linguistic behavior within a population and the application of computer science techniques to model this process. Lakkaraju’s primary research interests lie in developing computational models of behavior change in society such as attitude change and adoption of solar PV. More recently, he has focused on developing models that explore the link between social structure (social networks, roles/hierarchy) with cognitive structure (how concepts are interrelated, cognitive consistency, confirmation bias) with respect to problems of information dissemination and attitude change. He is currently developing a social experimentation platform (the Controlled, Large, Online Social Experimentation platform). He is co-principal investigator of a Department of Energy-funded project to develop data-driven models of residential PV adoption. Lakkaraju has more than 30 peer-reviewed publications in journals and conference proceedings.
Yevgeniy Vorobeychik is an assistant professor of Computer Science, Computer Engineering, and Biomedical Informatics at Vanderbilt University. Previously, he was a principal member of the technical staff at Sandia National Laboratories. Between 2008 and 2010 he was a postdoctoral research associate at the University of Pennsylvania in the Computer and Information Science department. Vorobeychik received Ph.D. and M.S.E. degrees in Computer Science and Engineering from the University of Michigan, and a B.S. degree in Computer Engineering from Northwestern University. His work focuses on adversarial reasoning in AI, computational game theory, security and privacy, network science, and agent-based modeling. He received a National Science Foundation CAREER award in 2017, was nominated for the 2008 ACM Doctoral Dissertation Award, and received honorable mention for the 2008 IFAAMAS Distinguished Dissertation Award. He was an invited early-career spotlight speaker at IJCAI 2016.
Michael Rossol is a member of the System Modeling & Geospatial Data Science Group in the Strategic Energy Analysis Center at NREL. His area of expertise is the processing and analysis of large data sets. Rossol received his Ph.D. in Materials Science from the University of California at Santa Barbara in 2015. His dissertation research focused on characterizing the structure and properties of advanced ceramic composites using digital image processing. Prior to joining NREL in the fall of 2016, Michael was a postdoctoral researcher at the University of Illinois at Urbana-Champaign, where he applied digital image processing to advanced polymers and polymer composites.