1

I have a scatter plot where most of the data points are near the origin. The closer to the origin, the more data points one can find. I have tried to use logarithmic scale for the $x$ and $y$ axes to display more data points near the origin. However, one problem is that there are a lot of data points on $x$ and $y$ axis. Using a logarithmic scale will not property display those points on $x$ and $y$ axes. I am wondering whether there are other way of scaling the $x$ and $y$ axes such that I can get more details near the origin, while still able to display the points on the $x$ and $y$ axes.

Note that all the data points range in $[0, 1]$ for both $x$ and $y$ coordinates.

enter image description here

KReiser
  • 65,137
andy90
  • 309
  • Test $x^\alpha$ and $y^\beta$ for some values of $\alpha,\beta$. – md2perpe Jun 23 '20 at 19:13
  • 1
    Stupid question: how sure are you that there are really points exactly on the x and y axis, and that this is not some artifact? I am asking because I saw many data processing pipelines where some x=max(x,0) appeared somewhere, because the value must be non-negative but is sometimes measured to be slightly negative... just wondering because 0 has measure zero in [0,1] but was hit several times... – NeitherNor Jun 23 '20 at 19:54
  • 1
    Do you know why the distribution is such ? That would be helpful. And do you know why there are zeroes ? –  Jun 24 '20 at 19:12
  • These 0s are precisely 0s. @Engineertryingmath – andy90 Jun 25 '20 at 00:42
  • 1
    What about using an inset plot in your current plot which represents the same data only zoomed in (e.g. x and y values between 0 and 0.1)? You can put it at the top- right of your current plot. You can also put a red box or so in the main plot showing the range if the inset plot...Anything else would only make sense after you answered @Yves question. – NeitherNor Jun 25 '20 at 05:46

0 Answers0