Making the Most of Data Mining and Visualization
Cheryl Pammer, MS
Senior Advisory Statistician
As the velocity, volume, and variety of data increase, many organizations find themselves drowning in data, yet starving for information. Six Sigma professionals have both the skills and the opportunity to bridge this information gap. As we add new analytical tools to our toolkit, we also need proven strategies to communicate results. In this session. I will use case studies based on my own experiences to demonstrate practical ways to pair advanced modeling techniques with modern visualizations so that we can communicate analytical results in more intuitive ways. Specifically, we will review how to apply machine learning techniques such as regression, logistic regression, decision trees, and clustering algorithms to mine observational process data. Then, we will use the model results to create unique visualizations that facilitate actionable insights. These visualization techniques, primarily based on model fitted values and predictions, will include binned scatterplots, parallel coordinates plots, dendrograms, bubble plots, 3D scatterplots, heat maps, and contour plots. We will discuss best practices related to visual art and design and then put these concepts into action by turning the results from statistical models into compelling visualizations that convey clear insights from the data and corresponding analytics. Finally, we will learn how to use visualizations to engage with executives, speak their language, and translate our data-driven insights into decisions and actions. • A review of machine learning techniques with specific applications of their value in mining observational process data. • Proven techniques for pairing advanced analytics with modern easy-to-understand visualizations.• Strategies for using model visualization to engage with executives and turn data-driven insights into decisions and actions. CLICK HERE to complete the session survey.