Bigger color-palette in matplotlib for SciPy's dendrogram (Python)

Question

I'm trying to expand my color_palette in either matplotlib or seaborn for use in scipy's dendrogram so it colors each cluster differently.

Currently, the color_palette only has a few colors so multiple clusters are getting mapped to the same color. I know there's like 16 million RGB colors, so...

How can I use more colors from that huge palette in this type of figure?

#!/usr/bin/python

from __future__ import print_function
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import colorsys
from scipy.cluster.hierarchy import dendrogram,linkage,fcluster
from scipy.spatial import distance
np.random.seed(0) #43984

#Dims
n,m = 10,1000

#DataFrame: rows = Samples, cols = Attributes
attributes = ["a" + str(j) for j in range(m)]
DF_data = pd.DataFrame(np.random.randn(n, m),#
                columns = attributes)

A_dist = distance.cdist(DF_data.as_matrix().T, DF_data.as_matrix().T)
DF_dist = pd.DataFrame(A_dist, index = attributes, columns = attributes)

#Linkage Matrix
Z = linkage(squareform(DF_dist.as_matrix()),method="average") #metric="euclidead" necessary since the input is a dissimilarity measure? 



#Create dendrogram
D_dendro = dendrogram(
         Z, 
         labels=DF_dist.index,
         no_plot=True,
         color_threshold=3.5,
         count_sort = "ascending",
         #link_color_func=lambda k: colors[k]
         )

#Display dendrogram
def plotTree(D_dendro):
    fig,ax = plt.subplots(figsize=(25, 10))
    icoord = np.array( D_dendro['icoord'] )
    dcoord = np.array( D_dendro['dcoord'] )
    color_list = np.array( D_dendro['color_list'] )
    x_min, x_max = icoord.min(), icoord.max()
    y_min, y_max = dcoord.min(), dcoord.max()

    for xs, ys, color in zip(icoord, dcoord, color_list):
        plt.plot(xs, ys,  color)
    plt.xlim( x_min-10, x_max + 0.1*abs(x_max) )
    plt.ylim( y_min, y_max + 0.1*abs(y_max) )
    plt.title("Dendrogram", fontsize=30)
    plt.xlabel("Clusters", fontsize=25)
    plt.ylabel("Distance", fontsize=25)
    plt.yticks(fontsize = 20)

    plt.show()

    return(fig,ax)
fig,ax = plotTree(D_dendro) #wrapper I made

#Dims
print(
    len(set(D_dendro["color_list"])), "^ # of colors from dendrogram",
    len(D_dendro["ivl"]), "^ # of labels",sep="\n")
# 7
# ^ # of colors from dendrogram
# 1000
# ^ # of labels

Answer 1

Most matplotlib colormaps will give you a value given a value between 0 and 1. For example,

import matplotlib.pyplot as plt
import numpy as np
print [plt.cm.Greens(i) for i in np.linspace(0, 1, 5)]

will print

[(0.9686274528503418, 0.98823529481887817, 0.96078431606292725, 1.0),
 (0.77922338878407194, 0.91323337695177864, 0.75180316742728737, 1.0),
 (0.45176470875740049, 0.76708959481295413, 0.46120723030146432, 1.0),
 (0.13402538141783546, 0.54232989970375511, 0.26828144368003398, 1.0),
 (0.0, 0.26666668057441711, 0.10588235408067703, 1.0)]

So you no longer need to be restricted to values provided to you. Just choose a colormap, and get a color from that colormap depending upon some fraction. For example, in your code, you could consider,

for xs, ys in zip(icoord, dcoord):
    color = plt.cm.Spectral( ys/6.0 )
    plt.plot(xs, ys,  color)

or something to that effect. I am unsure how exactly you want to display your colors, but I am sure you can modify your code very easily for achieving any color combinations you want ...

Another thing you can try is

N = D_dendro["color_list"]
colorList = [ plt.cm.Spectral( float(i)/(N-1) )  for i in range(N)]

and pass on that colorList.

Play around a bit ...