I'm using Python, and I have some data which can be projected on to points within an equilateral triangle whose side lengths sum to 1.
I'm not sure if there is an easy way to visualise a plot like this from Matplotlib or similar libraries, or if I'm just going to have to use a drawing package from scratch to make it happen. Any pointers gratefully recieved. Thanks!
If all you want to do is plot a few dots on a graph, you can infact use Matfplotlib's scatter plots for this:
https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.scatter.html
Using plt.xlim(*min*, *max*) and plt.ylim(*min*, *max*), you can set the limits of the graph manually to fit your values (might not be neccessary though).
You can draw lines on the graph for the triangle shape (if you need that): How to draw a line with matplotlib?
If you don't need a scale, you can even remove that to have a blank canvas: Matplotlib plots: removing axis, legends and white spaces
Related
I need to draw additional graphics on top of plotly go.Box traces, therefore I need to know X and Y coordinates for boxplot rectangle vertices. So far the only solution I came up with is basically recalculating everything (quartiles; X positions based on boxgap, boxgroupgap, etc.), then manually setting the y-axis range to know where everything will end up on the plot. This seems very cumbersome.
Is there a way in python to get the coordinates of go.Box boxplot elements, especially the grouped boxplots with categorical x-axis? As far as I understand these coordinates are calculated in JS frontend -- maybe there is some trick to get them back with Dash using callbacks?
I want to plot boxplots on top of the scattered points like this.
I know I have to bin the data into intervals first but I couldn't find the function that does all of this. Sample x and y data are saved here as .npy.
I would look into using matplotlib. Boxes can be drawn as such:
https://matplotlib.org/gallery/pyplots/boxplot_demo_pyplot.html?highlight=boxplot
and scatter plots can also be drawn as such: https://matplotlib.org/gallery/lines_bars_and_markers/scatter_demo2.html?highlight=scatter
There is a search functionality on their site, along with plenty of documentation on how to utilize their library.
As for your specific question, you can specify zorder when drawing many of the things in matplotlib, and you could use that to define your boxplots to be on top. I believe if no zorder is defined that it draws items in the order they are encountered in your program (so you could draw scatter plots and then box plots and they should appear correctly as in your diagram above!
Consider a Manhattan plot for a genome-wide association study. The density of dots at the bottom of the plot is very high -- individual points are no longer visible. I'd like to skip plotting the points that completely overlap with other points (even though their x,y is not identical) to reduce the plotting time and the size of the exported PDF. Any recipes for achieving this? Collision detection? Subsampling?
I'd like to use matplotlib, though this requirement is optional. Ideally, the output should be visually identical to the "full" plot.
Some background info on the plot type:
https://en.wikipedia.org/wiki/Manhattan_plot
In a standard 3D python plot, each data point is, by default, represented as a sphere in 3D. For the data I'm plotting, the z-axis is very sensitive, while the x and y axes are very general, so is there a way to make each point on the scatter plot spread out over the x and y direction as it normally would with, for example, s=500, but not spread at all along the z-axis? Ideally this would look like a set of stacked discs, rather than overlapping spheres.
Any ideas? I'm relatively new to python and I don't know if there's a way to make custom data points like this with a scatter plot.
I actually was able to do this using the matplotlib.patches library, creating a patch for every data point, and then making it whatever shape I wanted with the help of mpl_toolkits.mplot3d.art3d.
You might look for something called "jittering". Take a look at
Matplotlib: avoiding overlapping datapoints in a "scatter/dot/beeswarm" plot
It works by adding random noise to your data.
Another way might be to reduce the variance of the data on your z-axis (e.g. applying a log-function) or adjusting the scale. You could do that with ax.set_zscale("log"). It is documented here http://matplotlib.org/mpl_toolkits/mplot3d/api.html#mpl_toolkits.mplot3d.axes3d.Axes3D.set_zscale
I am plotting some scalar data as a contour plot with matplotlib.contourf. On top of it, I am plotting some vector data with matplotlib.arrow. The basic plot has come along OK, but now I need to put a box on the plot with a default-size arrow plus the data value to which it corresponds, so the viewer will know what kind of scale he is looking at. For instance, I need a box with a horizontal arrow of some length and, below that, some text like "10 cm/sec".
First, if anyone can give me a simple approach to this, I would be grateful.
Second, the approach I have tried is to do the contour plot, then plot the arrows, then add a rectangle to the plot like so:
rect=pl.Rectangle((300,70),15,15,fc='white')
pl.gca().add_patch(rect)
and then, finally, put my scale arrow and text on top of this rectangle.
This isn't working because the rectangle patch covers up the contour, but it doesn't cover up the arrows in the plot. Is there a way to move the patch completely "to the front" of everything else?
Got it. Using pylab.quiver and pylab.quiverkey functions. quiver produces a nice vector field with just a few lines of code, and quiverkey makes it easy to produce a scaling vector with text. And, for some reason, the arrows plotted with quiver are indeed covered by my rectangle, so it is easy to make the scaling arrow very visible. There are still some mysteries in all of this for me. If anyone wants to try to clear them up, would be much obliged. But I have a way now to do what I need in this instance.