Key Driver Analysis – Using CCR-Johnson’s, a more stable approach, able to handle all types of data
As market researchers, we are frequently asked to identify the key drivers of a market, including prescribing or purchase behaviour. As human beings, we are often unaware of exactly what leads us to make certain decisions, so to focus on stated importance is flawed. Instead, we need to statistically derive the importance of key drivers, by assess correlations between different data points. There are many excellent ways to do this, but we find that CCR-Johnson’s offers a number of unique features:
It produces more stable calculations across different samples, unlike conventional regression, making it more suitable for tracking.
It uses a unique cross validation algorithm to provide greater certainty as to which drivers are key in determining treatment choice
It offers a unique “stepping down” algorithm to select the optimal numbers of drivers
It can handle all types of data – binary, ordinal, multi-nominal, etc.
At Purdie Pascoe, we recently won ‘Analyst Team of the Year’ at the BOBI awards, in conjunction with our friends at The Stats People, for our analysis of the COVID-19 data both internationally and domestically. An element in our winning submission was Key Driver Analysis of case rates and deaths rates, using CCR-Johnson’s. Please click on the image below to see the dashboard which summarises our findings, including the case and death rates as well as the key drivers. Feel free to play with the dashboard by selecting different countries and different time frames, to see how they compare.
