Generating input data for a heat plot

[Tom Yates]

Dear friends,

I am trying to represent 3D data in 2D. The colours on the plot would be probabilities of TB infection and on the axes, I would have time (t) and ventilation rate (Q), keeping other variables fixed.

Probability = 1 - e^ (-Ipqt/Q) [the Wells Riley equation]

I am wondering about using the contour plot option in twoway, which I think I am comfortable with (there is a nice video on the STATA YouTube channel). I am struggling with setting up the input data.

I think I need a set of input data that look like this...

1, 1
1, 2
1, 3
[...]
1, 10000
2, 1
2, 2
2, 3
[...]
2, 10000
[etc, all the way up to...]
10000, 10000 [10000 hours and 10000 m3/hr]

The third column in the dataset would then be a simple formula derived from the first two columns.

I am not sure how well twoway would handle a 10000 x 10000 grid. I will probably have log scales on my contour plot so, ideally, would have more data points at lower values of t and Q and could afford to have a less dense grid at higher values of t and Q.

I am sure this is very simple but I am struggling!

Thanks,
Tom

[Scott Merryman]

I would recommend Ben Jann's -heatplot-. I think -twoway contour- with a million observations would take too long.

clear
set obs 1000
gen x = _n
expand 1000
bys x: gen y = _n
gen z = runiform() 
heatplot z x y

[Tom Yates]

Thanks, that is neat. I am unable to find a good help file for heatplot, though, and the twoway chart formatting options (scales, labels, etc) don't seem to work?

Best wishes,
Tom