Gradient colored scatterplot

[Fahad Mirza]

Hello everyone,

I am trying to make a scatter plot using the plottig scheme (or any scheme for that matter) that shows a change in color as we move from lower to higher values. Is there any way I can achieve this?

For example lower values take the color blue and as we go up the color transitions to red by fading into it.

Thank you

[Maarten Buis]

You could look at https://journals.sagepub.com/doi/pdf...867X0500500412
and more categories combine it with https://journals.sagepub.com/doi/pdf...867X1801800402

[ericmelse]

Dear Fahad,

You might benefit from Ben Jann's reply on the Stata 17 wishlist, working with his package heatplot
in response to my wish for Graded marker fill color by the values of a variable (like marker size), also Opacity controlled.

Best

[Fahad Mirza]

Maarten Buis thank you for the links however, mainly due to my lack of skill, I am unable to execute this task. Nonetheless very interesting color palettes and technique by Nick and Ben Jann

[Fahad Mirza]

ericmelse that sounds like a great idea

[Fahad Mirza]

I tried a method and it seemed to have worked

  
generate weight = var_1/1000           //This is just an example form of weight you can choose any other if you like
    
#delimit ;

heatplot weight var_1 var_2, discrete scatter
colors(blue red, opacity(60) ipolate(20))
p(mlc(black))
legend(off)
scheme(plotplain)

addplot(lfit var_1 var_2, lcolor("220 0 12") lpattern(dash))

title("Gradient Scatterplot with Line of best fit using Heatplot")

;
#delimit cr

[Fahad Mirza]

Thank you for suggesting heatplot, it really made a difference on how my graph looked. will be great if you all can add to this so that I can improve it.

[Justin Niakamal]

I think incorporating the size option would improve the graph.

[Fahad Mirza]

Justin Blasongame you mean linearly increase size of plots just like how colors are changing?

[Justin Niakamal]

Not quite. I mean to show some scale to the plots. Consider the example below, where the dots reflect the number of jobs -- this doesn't apply to your example, but it's an example I can think of where using this option helps.

* Example generated by -dataex-. To install: ssc install dataex
clear
input double local_emp float(wage lq)
 84279  160.9143  .6679937
 55280  42.59007 1.0586338
 25736 37.093792 1.0633941
 31394  51.12663  1.533422
 56712  47.90494  .7758535
 31578  47.97167 1.0501165
 73749  42.45197 1.0328938
 24839  49.11127 1.3231826
 22602  77.65968  .9109533
 82682  28.90806 1.0353069
 10415  154.1099 .56889135
125796  95.27658 1.1112541
457004 111.85567   .813385
149258 25.486544 1.0186863
 90220  24.87117 1.1197183
 34243  120.9658  .8878158
 43233  54.02937  .9697956
  2836 212.27405  .6684303
end


heatplot local_emp lq wage , discrete stat(asis) scatter size(local_emp)  ///
p(mlc(black)) colors( plasma , opacity(60) ipolate(20))    ///
ytitle("LQ", size(small)) ylab(, format(%5.2f)) ///
xtitle("Wage (\$000s)", size(small)) xlab(, format(%9.0fc)) ///
legend(off)    ///
scheme(plotplain)

[Fahad Mirza]

Justin Blasongame this looks good and yes this is making quite more sense plus the fact that now there are 3 information being presented. Thanks again for this.