Business intelligence under uncertainty       


 Lilli-Marie Pavka photo

Business intelligence under uncertainty - overview

User-centric business intelligence, multi-modal information and uncertainty

Business leaders and their teams face the challenge to take strategic decisions based on a vast amount of information on their business and its environment. At the same time, they have to take into account the uncertainty of future developments. Thus, high-quality decision making depends on three factors:

  1. the ability of a team to delineate the boundary between skilled intuitively-based decisions and potentially disastrous cognitive biases,
  2. the seamless integration of structured (e.g. Excel spreadsheets) and unstructured data (e.g. written text, graphs, PDF documents)into analytics tools to support, develop and challenge decisions,
  3. a careful framing of the uncertainties in parameters and driving factors that are most likely to affect how decisions will play out once implemented.

At IBM Research – Australia. we are leveraging IBM’s unique strengths in affective computing, machine learning, natural language processing, uncertainty quantification and analytics to create an IT platform for behaviourally-based strategic decision making under uncertainty that addresses each of the three key factors described. This platform will transform communication between data science teams and their clients and greatly reduce time to value.

One use case is a portfolio management tool based on a cognitive version of the growth rate vs. market share matrix. This matrix is commonly used to assist in invest/divest decisions. For instance, assets with low scores in both market share and growth rate are strong candidates for divestiture. In our tool, various sources of data such as spreadsheets, financial reports and text are used to locate business assets within the matrix and visualize uncertainties. The tool also leverages the Watson API Personality Insights to build individualized risk profiles of the decision makers. This information is used to make suggestions on likely cognitive biases at play as the decision-making process unfolds.