The problem with machine learning (ML) is that it gets talked about as if it were something fundamentally new for BI. Rather, ML is but the latest entrant in a storied line of analytic technologies, the purpose of which should be to improve the IQ of BI systems (and which should be subject to BI’s best practices). Thus, the ML discussion should begin in response to problems identified within the larger BI ecosystem that could be resolved through better analytics and automation based on analytics.
In this talk, Erik Thomsen will explain how ML compares to other widely deployed analytic technologies and will describe where it can add significant business value, e.g., for interpreting unstructured data, demand forecasting, and adaptive data fusion. He will then explain why it is crucial to systematically marry ML with real world semantics (knowledge or ontologies) both from the enterprise and from the “outside” world to improve confidence with individual ML applications, eliminate proliferation of ML siloes, and guarantee consistency across ML projects, thus improving the IQ of the overall BI system.