CODEWITHSUNDEEP
rgraphicstransform
Lloyd Christmas
Lloyd Christmas

Reputation: 1038

Transform to arbitrary coordinates to compare relative effect of inputs

I was originally going to post this on a math forum, but felt like the question was dependent on making a graphic, and the rules making graphics, so I ended up here. I use R for all my data work.

I have many different "types" that I want to compare against their limits (LL=lower limit, UL=upper limit). They all have different ranges and limits, but I want to compare how they change, relative to different inputs.

It's easier explained with examples. Here are my limits and targets for each type:

        LL   Target   UL
type1   30     33     36
type2   10     15     20

Here's some of my data points. I want to plot the effect of the "mode" on each, and compare them. The types are dependent on each other.

Mode   Meaure   value
1      type1    35
1      type1    36
2      type1    33
2      type1    32
1      type2    8
1      type2    9
2      type2    17
2      type2    18

The graph that I want to create will look like this:

graph

This would require mapping to a new, arbitrary coordinate system. I'm not sure how to do this conversion. I was thinking of going from -1 to 1. It would look something like this:

transform

My questions are:

  1. Is this an appropriate way to make a comparsion? (relay impact of "mode" on the different "type")
  2. How to do the transformation

Upvotes: 0

Views: 58

Answers (1)

Andy Reagan
Andy Reagan

Reputation: 639

Whether it makes sense to normalize the values for comparison of course depends on the what the value are (what the data is). If that makes sense, then like the commenter said, go ahead and normalize (...subtract lower limit, divide by the range of LL -> UL to put them in -1,1).

Encoding the categorical data using circles/squares is fine, you could use different hues as well.

Upvotes: 1

Related Questions