Recently I’ve been immersing myself in The Art Of Statistics by David Spiegelhalter – an excellent refresher on the principles of statistical analysis from the former head of the Royal Statistical Society.
In the introduction, Spiegelhalter makes an interesting observation on the failure of current statistics education, which focusses on “mathematical theory [rather] than understanding both why the formulae are being used, and the challenges that arise when trying to use data to answer questions.”
I thought this was an important point, and one that that applies equally to business intelligence reporting. A suite of new reporting tools have made preparing and visualising data easier than ever – however the foremost skill of any analyst is still understanding how these can be leveraged to answer questions.
“The needs of data science and data literacy demand a more problem-driven approach, in which the application of specific statistical tools is seen as just one component of a complete investigation cycle,” writes Spiegelhalter.
Instead of focussing just on the formulae that drive statistics, Spiegelhalter suggests a more holistic problem-solving model as the starting point of good analysis.
The PPDAC Model
The PPDAC model of analysis is borrowed from New Zealand’s school system, a world leader in statistical education.
It breaks down the process of any data driven inquiry into five distinct, and equally important stages:
- Problem: Understanding and defining the problem
- Plan: What to measure and how
- Data: Collection, management, and cleaning
- Analysis: Constructing graphs, looking for patterns
- Conclusions: Interpretation and new ideas for future analysis
By breaking down an analysis in this manner, the PPDAC model reminds us that defining a problem, ensuring data quality, and considering the real world outcomes of your analysis are as important as the technical components of reporting (data prep and charts).
“Although in practice the PPDAC cycle… may not be followed precisely, it underscores that formal techniques for statistical analysis play only one part in the work of a statistician or data scientist,” writes Spiegelhalter.