9
votes
Favorite data visualization toolset?
I'm primarily a non-programmer these days, but have a fairly extensive background in statistical analysis - seeking recommendations for best/cheapest/easiest-to-learn data visualization tools. I have access to PowerBI and Tableau through work, but any other recommendations are welcome. You can take the SQL-family relational database query skills for granted, but not necessarily noSQL, Hadoop or the other popular big data sources.
As someone who's job is to work with data visualization toolsets and do data anlaysis, I can't recommend Tableau enough. There's a reason that Tableau has won the gartner magic quadrant for basically the last decade. It's the best, hands down. Every other tool set in the last 2-3 years has been slowly copying their model (and might I say, poorly) because they've realized that they can't compete without imitation.
I was an intern doing some data analysis for a few months. Tableau was easy for me to pick up and ridiculously powerful - I never even came close to using even half of its capabilities. @Gaywallet's recommendation is spot on.
If I could ask you analysis guys a question. I know very little about the tools sets out there. Is there any need for a non-convex 3d hull turned into a triangle mesh? In other words, say you started with a random distribution of points in 3d space, literally a cloud of points. Then performed a series non-convex (concave) hulls. Basically drawing a circle around the points on a plane so as to contain all points within a minimum size. Each plane would be set at a different angle, like segments in an orange. Once done, you calculate the required triangles to fit the segments. All points would be joined together in vertice based (kind of) sphere. Part of the calculation would be an Epsilon value that would determine whether a set of neighboring points formed a solid wall or contained a gap between them. By varying this Epsilon value, tunnels would appear within the hull that you could traverse through. The bigger the Epsilon, the more solid the final object.
Does anyone use or need a visualization tool like this?
Outside of very specific applications (perhaps engineering), new ways to visualize data are not needed. Generally data visualization is used to aid in analysis and analysis asks questions like "how can I make x process more efficient?" or "is y a good investment?" or "help me understand the waste in process z". The vast majority of data analysis visualization is comparing something over time (line chart), comparing between categories (bar chart), or presenting raw data for detailed dive (tables).
I'm an R programmer, so I'm obviously biased towards ggplot2. I understand there's a python clone as well, and I recently found G2, which is a grammar of graphics implementation in JavaScript that looks promising (most of the documentation is in a language I don't read, but Google Translate did well enough that I could understand the code comments).
I'm not a Plot.ly user, but I have seen enough of their stuff to think it looks cool, even if isn't totally open source. Not a Tableau fan, personally, but that's just me being a curmudgeon and preferring open-source tools.
I won't lie and say that R is easy to learn, but with RStudio, the learning curve isn't as steep as it used to be. Last week I taught a couple of HS students the basics of R in about 2 hours, and the basics of ggplot2 in about the same length of time. Now they're using RStudio's gui to import their own data and are successfully playing around creating their own plots. Neither one has programmed before, so anyone who is familiar with data manipulation or programming would probably be able to pick things up even faster.
I'm old-school: gnuplot. Been meaning to learn R or play with d3g.