Machine Learning and Big Data

Abstract

The last 10 years saw a remarkable increase of data available to marketers. What was considered 5 years ago as big data (see e.g., Vol. I, Sect. 3.5.6) was based on hundreds of thousands of observations (e.g., Reimer et al. 2014). Today this is considered by data scientists as an average sample size. Consumer handscan panels and regular brand tracking allowed us to build models combining consumer actions with mindset metrics (e.g., Hanssens et al. 2014; Van Heerde et al. 2008). And of course, click stream data inspired a brave new world of modeling (prospective) customers’ online decision journeys (Bucklin and Sismeiro 2003; Pauwels and Van Ewijk 2014). Within the last two years we see social media interaction data coming more and more into marketing research (see e.g., Borah and Tellis 2016; Ilhan et al. 2016). Through this source marketers get access to customer and user generated content. Especially service and fast mover consumer goods as entertainment brands enjoy a high volume of interactions (see e.g., Henning-Thurau et al. 2014) in different social media channels that is linked with the company’s future sales performance. Social media again boosts the number of data available to marketers. Instead of now facing hundreds of thousands of observations, researchers are very likely to encounter millions of comments, likes and shares in even short observation periods. Combining this new data with existing sources, will provide new opportunities and further develop marketing research (Sudhir 2016). Verhoef et al. (2016) provide many examples how big data (analytics) can be used to create value for customers and firms.