About

Michael COnklin

I started building statistical models in my first job at General Electric.  For the past 30+ years I have led advanced analytics teams at GfK (Now NIQ) and MarketTools (now MetrixLab). I have constantly slaked my thirst for knowlege by expanding the repetoire of models used to solve marketing and research problems.  Current favorites include:

• Shapley Value Analysis - applying game theory to product line optimization.

• Bayesian Networks to identify the interrelationships between drivers that are often masked in a standard regression model.
• Data Fusion - creating a combined dataset from two different datasets with a small number of items in common.
• Vector Autoregression time series models to identify relationships between several separate time series.
• Customer satisfaction models that make clear that the drivers of high satisfaction are likely different from the drivers of dissatisfaction.
• Random Forest models to find key interactions between variables.