By David Steinberg.
MIS Quarterly is a management and information science journal that publishes many articles that make interesting and unusual use of data. A good example is the forthcoming article by Geva et al., which exploits internet data for sales prediction. Their work builds on many previous articles that use data from social media websites to predict offline economic outcomes such as sales. However, they also point to many limitations in social media value: it may be unrepresentative, inducing biases, and even intentionally manipulated. An alternative option is to use data from search engine logs for prediction. Although free from the above weaknesses, these data also have limitations, for example that it is impossible to know if a search has resulted from a positive or negative sentiment toward the product. This article is the first to combine these sources of data for the purpose of prediction. Specifically, the authors look at the prediction of automobile sales in the US. The social media data are extracted from Google’s comprehensive index of Internet discussion forums; the search engine log data come from Google search trend data. The authors find that significantly better predictions result when combining the two types of data than when using either one or the other alone. An interesting finding is that the improvement in predictive power is especially strong for “value” care brands, but for “premium” brands, the improvement is less pronounced.
Read the paper:
Using Forum and Search Data for Sales Prediction of High-Involvement Projects. Geva, T., Oestreicher-Singer, G., Efron, N. and Shimshoni, Y. Forthcoming Articles for MIS Quarterly.