Python scatter plot: soft clustering

Question

I have 2D data that I clustered using EM algorithm with soft classification. There are 3 different clusters, therefore I have a probability vector with dimension (n_clusters, n_datapoints).

Now I'd like to plot the individual datapoints in a scatter plot and assign a certain color to each cluster. The color of each point is given according to the probability to be in each cluster and thus a mixture of the cluster colors.

All I could achieve by now is the following with red, green and blue cluster colors

Scatter plot

by using the following lines of code:

for n in range(X.shape[0]):
    color = np.array([P[0,n],P[1,n],P[2,n]])[np.newaxis]
    plt.scatter(X[n,0],X[n,1],c=color)

How can I assign a different, specific color to each cluster? E.g. orange for class 0, blue for class 1, magenta for class 2.

Answer 1

I would create a dict where each cluster has it's own color. Then I would just add all cluster-colors together multiplyed by their probability:

colors = []
cluster_colors = {0:np.array([255,0,0]),1:np.array([0,255,0]),2:np.array([0,0,255])}
for n in range(X.shape[0]):
    color = np.zeros([3])
    for c in range(P.shape[0]):
        color += cluster_colors[c]*P[c,n]

    colors.append(color)

When I understood your data right, this should just work.

Answer 2

You can use plotly which is a good charting library outthere for this purposes.An example will be like below where you can set size and colour for each category.

#import plotly library 
import plotly.express as px
#get the iris datset 
df = px.data.iris()
#plot the scatter plot by setting size and colour 
fig = px.scatter(df, x="sepal_width", y="sepal_length", color="species",
                 size='petal_length', hover_data=['petal_width'])
fig.show()