I am starting to use 3D graphs to visualize relationships in my data. The reason I am doing this is to try and maximize my predictor variables. Since I am working with financial data the relationships are very weak. I am hoping that using 3D graphs will help me find groups and help me with deciding on transformations to predictor variables.

I have created a 3D graph that has 3 predictor variables. I have then coloured by my dependent variable which is a binary event.

Since I am new to 3D modelling I am curious to know what some of you see in this data (if anything at all). Because it is Christmas tomorrow, I have decided to colour the dependent varibale red and green. :)

Note* the dependent variable is actually a continuous number however I have filtered the data.frame to only include the bottom quartile and the top quartile. Which is now green for bottom and red for top. Also note that there are equal amounts of red and green counters

Edit* James Phillips asked for a link to the graph. I put together a shiny app that is hosting this 3D plot. Here is a link to the app. https://financialtrading.shinyapps.io/crossvalidated/

enter image description here enter image description here enter image description here enter image description here

  • $\begingroup$ A) The maker size is a bit too large, such that the points farthest away from the viewer are being obscured - this can interfere with pattern detection. I suggest reducing the point marker size for this reason. B) Would you please post or link to the plotted data? $\endgroup$ Dec 24, 2019 at 12:48
  • $\begingroup$ Hey james I edited the original comment to include a shiny app with the graph $\endgroup$ Dec 24, 2019 at 17:51
  • $\begingroup$ Can you post the data for people to play with. Briefly, I wouldn't do this, but I'm sure there's a better way. $\endgroup$ Jan 9, 2020 at 19:59


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.