Matplotlib Colorbar 与散布颜色不同？

Question

The code below takes the data (sample is copied below), and performs a scatter plot where the shape of the scattered point depends on the string value of the first column of the data. the shapes look correct, but the Colorbar does not correspond the normalized colors of the scattered points! Calling directly the Colorbar makes it independent from the plot, while calling it inside the loop will only show it multiple times... So, the Colorbars need to be independent (fake), but calibrated using the same data. The part I'm not sure of is : c=cmap.to_rgba(i + 1) The final image is attached (img.png)

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl

ax = plt.gca()

df = pd.read_csv('data.txt', delimiter="\t")
df.columns = ["type", "bv", "ron", "fom"]
df = df._convert(numeric=True)

norm = mpl.colors.Normalize(vmin=df.fom.min(), vmax=df.fom.max())
cmap = mpl.cm.ScalarMappable(norm=norm, cmap=mpl.cm.hot)
cmap.set_array([])

for i in range(len(df.type)):
    if df.type[i] == 'a':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='o', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )
    if df.type[i] == 'b':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='d', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )
    if df.type[i] == 'c':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='h', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )
    if df.type[i] == 'd':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='H', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )
    if df.type[i] == 'e':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='s', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )
    if df.type[i] == 'u':
        sc = ax.scatter(df.bv[i], df.ron[i], marker='<', edgecolors='black', alpha=0.8, s=100, c=cmap.to_rgba(i + 1),
                        )

plt.yscale('log')
plt.xscale('log')
plt.grid(True, which="both", ls="-", alpha=0.1)
plt.colorbar(sc, ax=ax, norm=mpl.colors.Normalize(min(df.fom), max(df.fom)), cmap='hot', alpha=0.8)
plt.show()

Image:

final image

Data sample:

type    ron bv  fom
b   23  57  141,2608696
c   3238    535 88,39561458
d   11000   858 66,924
b   115 35,9    11,20704348
b   28  28  28
a   5   23  105,8
d   14500   977 65,82958621
d   3090    477 73,63398058
e   94  50  26,59574468
e   53  127 304,3207547
b   32,4    35,2    38,24197531
e   7,8 25  80,12820513
c   57  75  98,68421053
c   91  100 109,8901099
b   49  55  61,73469388
b   95  82  70,77894737
u   7,42    22,48   68,10652291

Answer 1

Try restructuring your scatter plot as:

cmap = mpl.cm.hot
for ...
    sc = ax.scatter(df.bv[i], df.ron[i], marker='<', edgecolors='black',
                    alpha=0.8, s=100, c=df.fom[i], norm=norm, cmap=cmap)
    ....
fig.colorbar(sc)  # without the norm and cmap and call outside the loop