How to smooth data points in scatter plots in python?

Question

I have a scatter plot where I am coloring each data point based on an array:

 plt.scatter(xs,ys,c=av,cmap=plt.cm.hot,s=50,alpha=0.5)

In [96]: xs.shape
Out[96]: (5594,)

In [97]: ys.shape
Out[97]: (5594,)

In [98]: av.shape
Out[98]: (5594,)

and this is the result:

Now, I want to keep the color but smooth the data points to get a smoothed scatter plot, something like this (from this post) or this image:

Comment: I figured that if I can add more points to my xs, ys, zs I can make the scatter plot with more data points, hence, it will look more like a heatmap plot, which is what I want. Now, for every point in xs, ys, zs, I want to add additional points with similar values around original points. Ideally, these additional points should form a normal distribution around actual original points in the xs, ys, zs. Is there a statistical tools to do this task? E.g. How to change [1, 5, 10] to [0.9,0.98,1,1.02,1.1, 4.9,4.98,5,5.02,5.1, 9.9,9.98,10,10.02,10.1] ?

Isn't that what `alpha` does? You can also add `edgecolors='none'` to remove your edges and make it smoother — CAPSLOCK, Mar 22 '19 at 15:32
The post you link to directly shows some ways to do that. What exactly is the problem? — ImportanceOfBeingErnest, Mar 22 '19 at 15:36
I want to use Gaussian kernel densities to smooth out data points. The post that I am linking is doing this but I also want to control the colormap. I have an array for color of each data point. — A.Razavi, Mar 22 '19 at 15:40

CAPSLOCK · Answer 1 · 2019-03-28T08:52:29.053

0

OP: Is there a statistical tools to do this task? E.g. How to change [1, 5, 10] to [0.9,0.98,1,1.02,1.1, 4.9,4.98,5,5.02,5.1, 9.9,9.98,10,10.02,10.1] ?

obs=[1, 5 ,10]

syntheticobs=np.random.normal(0,0.1,(6,3))+obs
synthethicobs

Out[]:array([[ 1.02166209,  5.00716569,  9.96726293],
             [ 0.96727493,  4.94823697, 10.03424305],
             [ 1.10756036,  5.12464335,  9.86776082],
             [ 0.97866246,  5.12743117, 10.06647638],
             [ 0.87842188,  5.00994338, 10.1114983 ],
             [ 1.10728294,  4.82523615, 10.03642462]])

edited Mar 28 '19 at 08:52

answered Mar 22 '19 at 15:34

CAPSLOCK

6,243
3
33
56

Please expand your answer to include a detailed explanation on how this solves the issue – Sterling Archer Mar 22 '19 at 16:24
Well removing the edge will give a smoother feeling to the picture but imho the question isn't really clear.. My guess is either this or OP wants to reduce the "spread" of the colormap. I'm not sure why anyone would want that thou – CAPSLOCK Mar 22 '19 at 21:53

How to smooth data points in scatter plots in python?

1 Answers1