[Bug]: legend and actual colors in plot are inverted

[JosephBARBIERDARNAL]

Bug summary

I'm not sure if it's a bug or just me not using the functions well enough, but I think that when I add a colour bar or create a legend, the colors are "inverted" (see the plots below) and therefore unreliable.

Code for reproduction

import matplotlib.cm as cm
import geopandas as gpd
import matplotlib.pyplot as plt
from matplotlib.patches import Patch
from mpl_toolkits.axes_grid1.inset_locator import inset_axes

data = gpd.read_file('https://raw.githubusercontent.com/JosephBARBIERDARNAL/misc-dataviz/main/country-co2/data.geojson')
# load the colormap
cmap = cm.Greens
background_color = 'white'
text_color = 'black'

# legend values
value_ranges = [0, 2, 4, 6, 8, 10, 12]
labels = ['0-2 t', '2-4 t', '4-6 t', '6-8 t', '8-10 t', '10-12 t', '12+ t']
def create_legend_elements(cmap, value_ranges, labels):
    legend_elements = []
    n = len(value_ranges)
    for i in range(n-1):
        value = (value_ranges[i] + value_ranges[i+1]) / 2 / value_ranges[-1]  # Normalize the value to [0, 1]
        legend_elements.append(Patch(facecolor=cmap(value), label=labels[i]))
    return legend_elements
legend_elements = create_legend_elements(cmap, value_ranges, labels)

# initialize the figure
fig, ax = plt.subplots(figsize=(10, 10), dpi=300)
fig.set_facecolor(background_color)

# create the plot
data.plot(ax=ax, cmap=cmap, edgecolor='black', linewidth=0.5)

# annot norway and sweden
norway = data.loc[data['country'] == 'Norway']['co2Capita'].values[0]
sweden = data.loc[data['country'] == 'Sweden']['co2Capita'].values[0]
fig.text(0.35, 0.67, f'Norway: {norway}', fontsize=8, fontweight='bold', color='black')
fig.text(0.45, 0.62, f'Sweden: {sweden}', fontsize=8, fontweight='bold', color='black')

# custom axis
ax.set_xlim(-11, 41)
ax.set_ylim(32, 73)
ax.set_axis_off()

# add the patches to the plot area as a legend
rectangle_width = 2
rectangle_height = 1.5
legend_x = 35
legend_y_start = 65
legend_y_step = 1.5
for i, element in enumerate(legend_elements):
    ax.add_patch(plt.Rectangle((legend_x, legend_y_start - i * legend_y_step), rectangle_width, rectangle_height,
                               color=element.get_facecolor(), ec='black'))
    ax.text(legend_x + 2.5, legend_y_start - i * legend_y_step + 0.7, element.get_label(), 
            fontsize=12, color=text_color, va='center')
    
# add colorbar
cax = inset_axes(ax, width='5%', height='50%', loc='center left')
sm = plt.cm.ScalarMappable(cmap=cmap, norm=plt.Normalize(vmin=0, vmax=12))
sm._A = []
cbar = plt.colorbar(sm, cax=cax)
                 
plt.show()

Actual outcome

Expected outcome

Additional information

don't know any fix or why is that behaviour happening.

Operating system

macOS sonoma 14.5

Matplotlib Version

3.9.0

Matplotlib Backend

module://matplotlib_inline.backend_inline

Python version

3.10.5

Jupyter version

No response

Installation

pip

[timhoffm]

Seems correct to me. The colormap Greens goes from light (0) to dark (1)

Your drawn colorbar from sm = plt.cm.ScalarMappable(cmap=cmap, norm=plt.Normalize(vmin=0, vmax=12)) is consistent with that.

[JosephBARBIERDARNAL]

In my example I displayed Sweden and Norway values to illustrate that lower values are darker in the map. However both colorbars display the opposite.

[rcomer]

I think you may have more luck asking this question over at geopandas, since geopandas drew the map.

[JosephBARBIERDARNAL]

I will try to reproduce it using only matplotlib but yes you're right!

[JosephBARBIERDARNAL]

since the following code:

import matplotlib.pyplot as plt
from matplotlib.patches import Patch
import numpy as np
import pandas as pd

# Generate fake data
np.random.seed(0)
data = {
    'Country': ['Norway', 'Sweden', 'Denmark', 'Finland', 'Iceland', 'Germany', 'France', 'Italy', 'Spain', 'Portugal'],
    'CO2 per Capita': np.random.uniform(0, 12, 10)
}
df = pd.DataFrame(data)

# Legend values
value_ranges = [0, 2, 4, 6, 8, 10, 12]
labels = ['0-2 t', '2-4 t', '4-6 t', '6-8 t', '8-10 t', '10-12 t', '12+ t']
cmap = plt.cm.Greens

def create_legend_elements(cmap, value_ranges, labels):
    legend_elements = []
    n = len(value_ranges)
    for i in range(n-1):
        value = (value_ranges[i] + value_ranges[i+1]) / 2 / value_ranges[-1]  # Normalize the value to [0, 1]
        legend_elements.append(Patch(facecolor=cmap(value), label=labels[i]))
    return legend_elements

legend_elements = create_legend_elements(cmap, value_ranges, labels)

# Initialize the figure
fig, ax = plt.subplots(figsize=(10, 10), dpi=300)
fig.set_facecolor('white')

# Create the plot
bars = ax.bar(df['Country'], df['CO2 per Capita'], color=cmap(df['CO2 per Capita'] / value_ranges[-1]), edgecolor='black')

# Annotate Norway and Sweden
norway = df[df['Country'] == 'Norway']['CO2 per Capita'].values[0]
sweden = df[df['Country'] == 'Sweden']['CO2 per Capita'].values[0]
ax.text(0, norway + 0.5, f'Norway: {norway:.2f}', fontsize=8, fontweight='bold', color='black')
ax.text(1, sweden + 0.5, f'Sweden: {sweden:.2f}', fontsize=8, fontweight='bold', color='black')

# Custom axis
ax.set_xlabel('Country')
ax.set_ylabel('CO2 per Capita (t)')
ax.set_title('CO2 Emissions per Capita by Country')
ax.set_facecolor('white')

# Add the patches to the plot area as a legend
rectangle_width = 0.2
rectangle_height = 0.5
legend_x = 10
legend_y_start = 10
legend_y_step = 1.5
for i, element in enumerate(legend_elements):
    ax.add_patch(plt.Rectangle((legend_x, legend_y_start - i * legend_y_step), rectangle_width, rectangle_height,
                               color=element.get_facecolor(), ec='black'))
    ax.text(legend_x + 0.3, legend_y_start - i * legend_y_step + 0.25, element.get_label(), 
            fontsize=12, color='black', va='center')

# Add colorbar
sm = plt.cm.ScalarMappable(cmap=cmap, norm=plt.Normalize(vmin=0, vmax=12))
sm._A = []
cbar = plt.colorbar(sm, ax=ax, orientation='vertical')
cbar.set_label('CO2 per Capita (t)')

plt.show()

works as expected, I will redirect my issue to geopandas. I don't the issue comes from matplot

I'll let you close the issue if that's okay.

[nicholas-ys-tan]

Issue was due to not specifying column name in GeoDataFrame.plot() hence uses the index for shading instead of values - the dataframe was sorted in descending order of the values OP wanted to plot, giving the unfortunate perception that the colour scheme was inverted.

Details here: geopandas/geopandas#3323 (comment)