That looks like it might be exactly what I'm hoping to do. Coming from a 
gnuplot background, I'm having a lot of trouble with the Gadlfy syntax, 
though. I can't seem to find any comprehensive documentation that lists all 
of the commands/options, and trying to understand it inductively from the 
minimal examples on the Gadfly site is frustrating. So far I haven't been 
able to make the contour plot that I'm hoping to make. Is there a resource 
you would recommend for learning Gadfly?

On Monday, May 16, 2016 at 11:56:14 AM UTC-4, Tom Breloff wrote:
>
> If your data is gridded (but just in vector form), then likely you could 
> just "zmat = reshape(z, numrows, numcols)" and then "contour(zmat)".
>
> On Mon, May 16, 2016 at 11:49 AM, <[email protected] <javascript:>> 
> wrote:
>
>> Yes, my data is gridded. And no, I don't plan to sample these density 
>> values later. I just want to plot it and see what it looks like :) I'm not 
>> sure how to reshape the data into a 2D array though, or how to make a 
>> contour plot from a dataset rather than from a function. Is there anything 
>> on this in the documentation? 
>>
>> Thanks for your help, 
>>
>> Amelia
>>
>>
>> On Friday, May 13, 2016 at 8:30:36 PM UTC-4, Scott T wrote:
>>>
>>> Two key questions - is your data gridded? And do you plan to sample from 
>>> these density values later, or are you just wanting to plot it and see what 
>>> it looks like?
>>>
>>> If your data is gridded (your ~10000 lines cover every combination of x 
>>> and y values in the range that you are interested in), then you can use the 
>>> contour command in Gadfly, which is the volcano plot you described. You'll 
>>> first need to reshape the data so it's a 2D array: think of it as 
>>> displaying a 2D image, where the number at each point is the density. 
>>> However, for displaying this kind of data, I prefer heatmaps, and I don't 
>>> know if Gadfly supports those - you may have to look into the histogram2d 
>>> command.
>>>
>>> If it is not gridded (the x and y points don't have any particular 
>>> structure to them), it's still possible, but you have to choose a way to 
>>> decide how you want to turn it from unstructured data into a 2D image. The 
>>> histogram2d approach that Tom showed above is one option, where you treat 
>>> each density measurement as a weighted measurement in a histogram. But if 
>>> your data represents single measurements of a function that has meaningful 
>>> values away from those measured points, you probably want to interpolate 
>>> between those points. For this you can use a package like Dierckx, which 
>>> does interpolations on unstructured data. I also have some simple code that 
>>> does barycentric triangular interpolation between unstructured points, in 
>>> case you wanted to have a look at that. 
>>>
>>> This may be overkill, however, if you just want to look at the data and 
>>> don't plan to interpolate or draw from those density values later. If 
>>> that's the case, the trisurface plot above might be just what you need for 
>>> showing you the shape of your density data.
>>>
>>> Whatever you choose, I can recommend Tom's Plots package as a nice 
>>> interface to the other plotting packages in Julia - it makes it easy to 
>>> switch between different plotting options like Gadfly and PyPlot depending 
>>> on what features they offer.
>>>
>>> Cheers,
>>> Scott
>>>
>>> On Friday, 13 May 2016 15:34:19 UTC+1, [email protected] wrote:
>>>>
>>>> Dear Julia users, 
>>>>
>>>> I have a rookie question about plotting in Gadfly. I have some density 
>>>> data in a plain-text file in the form  of x y d, where d is the density at 
>>>> the point (x,y). I have about 10,000 lines of this data. I'm currently 
>>>> plotting old-school using gnuplot and since I don't like the looks of what 
>>>> I've been able to make, I'm hoping to be able to do something more elegant 
>>>> like Gadfly. I'm a relatively new Julia user as well. I like the "volcano" 
>>>> contour plot from the Gadfly documentation ( second plot from the top at 
>>>> http://dcjones.github.io/Gadfly.jl/geom_contour.html). I'm just not 
>>>> sure how to go about it. 
>>>>
>>>> Has anyone done something like this before? I think it could be a 
>>>> really beautiful way to represent my data if I can get it to work. Any 
>>>> hints or suggestions would be greatly appreciated!
>>>>
>>>> Cheers, 
>>>>
>>>> Amelia
>>>>
>>>
>

Reply via email to