EDO

 

Exploratory data analysis for complex models

Exploratory data analysis (EDO) is one of the essential steps when building a data science pipeline. Nowadays, there are many technology choices every step of the way. When dealing with graphics and data visualization, libraries and tools are abundant. However, theories of statistical graphics are rather in-depth and tend to prescribe the most appropriate approaches. I like the summary in section 3 of this paper. The author, Andrew Gelman, discusses the “universal grammar” of data visualizations and points out research into methods that seem to prevent being “perversely wasteful of data.” Even though some of the citations are from the pre-python era, there are valid studies on psychological models of cognition and visual perception. These models should still be accurate for OLED displays and 144 Hz gaming monitors.  

http://www.stat.columbia.edu/~gelman/research/published/edafinal.pdf