No application is required for this workshop.
We kindly ask that if you plan to attend the workshop that you register for the workshop between August 15, 2018 and September 1, 2018.
In order to ensure a high quality experience, registration for workshops will be limited. Workshop registration is available on a first-come, first-serve basis. Walk-ins are welcomed, only if space is available.
Saturday Workshops at the Annual Conference are included in the registration for the Annual Conference. Participants in workshops must be registered for the Annual Conference.
sponsored by the Research Methods Community
Machine learning (ML) is rapidly changing society and research. This workshop provides a practical ML introduction for strategy scholars without ML experience. Participants will get direct exposure to three key ML topics: regularization, natural language processing, and clustering. Under the guidance of experts, participants will analyze data in computer labs and learn about ML firsthand. This workshop also provides participants with information on how to continue learning how to apply ML after the workshop and motivation to incorporate ML in their future research.
The topics are chosen to showcase the ML way of thinking and to connect to strategy research. The first topic is regularization. ML builds models from data. The main risk is that the model is too specific to a sample, i.e. the model has been overfitted. The model explains well this sample but not another. A key ML technique to prevent overfitting is regularization, i.e. making a model less specific to a given sample to predict better another sample. An example of regularization is the least absolute shrinkage and selection operator (LASSO).
The second topic is natural language processing (NLP). While there is a long tradition of NLP in computational linguistics, in the ML context, we are referring to the more recent classes of techniques such as vector-space models, topic models, and other deep learning models that tend to not rely much on ex-ante specified linguistic rules, but rather use ML techniques to detect patterns in the text, and infer higher-order conceptual structures from it. An example of NLP is sentiment analysis.
The third topic is clustering. ML distinguishes between supervised and unsupervised learning. A dependent variable is present in supervised learning and absent in unsupervised learning. Clustering is a technique for unsupervised learning. It finds patterns in the data based on similarity of the observations. An example of clustering is K-means clustering.
We include also a presentation on ML and publishing that will address the challenges of implementing and publishing innovative research methods.
Machine learning (ML) is rapidly changing society and research. Societally, ML increasingly influences the job interviews we get invited to, the products and services we get offered, the prices we need to pay, the political messages we see, and even the potential romantic partners we are introduced to. As for research, ML allows for the analysis of new data (more (e.g. “big” data) or different (e.g. text, images, voice, video) data, introduces new methods (e.g. LASSO, random forests, neural networks), influences new practices (e.g. systematic variable selection, hold-out samples for theory testing, combinations with existing econometric tools to strengthen casual inference (e.g. instrumental variable regression)), and will hopefully help find new answers to existing research questions as well as open up new research avenues.
Quite literally, ML is about getting computers to learn without explicitly programming them (Samuel, 1959). A typical ML method detects patterns in observed data, builds a model, and then makes predictions in unseen data. ML originated in computer science and statistics and is foundational to artificial intelligence and data science. It is one of today’s most rapidly growing technical fields (Jordan and Mitchell, 2015) with adoptions in the social sciences, including economics (e.g. Mullainathan and Spiess, 2017), and marketing (e.g. Cui, Wong, and Lui, 2006). Applications within the strategy field are still limited and often in working papers (Hannan et al. 2017; Vanneste and Gulati, 2018; Furman and Teodoridis, 2018; Menon, Choi, and Tabakovic, 2018). However, the possibilities opened up by these techniques are particularly attractive here given that a significant amount of strategically relevant information is buried deep in text, and that the phenomena studied are highly complex and context dependent – situations where the power of ML truly comes to the fore. Thus, the question is not whether we will be doing ML, but how.
The workshop is designed as an introduction to ML for scholars without ML experience. The focus of the workshop is on active learning. The workshop discusses three topics: regularization, natural language processing, and clustering. The structure for a given topic is as follows. First, an expert introduces a specific ML topic. Second, participants analyze data on their laptops. Data and scripts will be provided. During the computer lab, the experts and organizers will facilitate and provide practical assistance. Third, the expert concludes a topic by reflecting on the results. Table 1 shows the timing of the workshop.
|10||Introduction - What is ML?|
|45||Topic 1: Regularization - Computer Lab|
|45||Topic 2: Natural Language Processing - Computer Lab|
|45||Topic 3: Clustering - Computer Lab|
|10||ML and Publishing|
|10||Outro - How can I learn more about ML?|
Cui, G., Wong, M. L., & Lui, H.-K. (2006). Machine learning for direct marketing response models: Bayesian networks with evolutionary programming. Management Science, 52(4): 597–612.
Furman, J., & Teodoridis, F. (2018). The cost of research tools and the direction of innovation: Evidence from computer science and electrical engineering", Working paper
Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends, perspectives, and prospects. Science, 349(6245): 255–260.
Hannan, M, Le Mens, G., Hsu, G., Kovacs, B., Negro, G. , Polos, L., Pontikes, E., & Sharkey, A. 2017. Concepts and Categories: Foundations for Sociological Analysis. Book (under review
) Menon, A. R., Choi, J., & Tabakovic, H. (2018). What You Say Your Strategy Is and Why It Matters: Applying Machine Learning to Unstructured Text in Strategic Management. Working paper
Mullainathan, S., & Spiess, J. (2017). Machine learning: An applied econometric approach. Journal of Economic Perspectives, 31(2): 87–106.
Samuel, A. (1959). Some studies in machine learning using the game of checkers. IBM Journal of Research and Development, 3(3): 210–229.
Vanneste, B. S., & Gulati, R. (2018). Generalized trust, external sourcing, and performance. Working paper
The workshop is designed for a broad audience and will be particularly relevant for those interested in applying ML methods, or anyone who simply wants to know more about ML and its potential for strategy research. This is an introductory ML workshop. Hence, no ML experience is assumed.