带有颜色栏和日期时间轴刻度的散点图

Question

I am getting lost in different methods used in matplotlib.

I want to create a colour-coded scatter plot with a colorbar on the side and datetime on the x axis.

But depending on how I define my ax , I get different errors. Below is the core of my code:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import matplotlib.cm as cm
import matplotlib.dates as mdates

#.....loading files etc.

norm = mcolors.Normalize(vmin=0,vmax=1000)
timerange = pd.date_range(start='2015-01-01', end='2016-01-01', freq='30D')

### PLOTTING 
fig = plt.figure(figsize=(6.,5))
ax = fig.add_subplot(111)

for Af in Afiles:
    for index, row in Af.iterrows():
        time = pd.to_datetime(row['date'], format="%Y-%m-%d")
        plt.scatter(time, row['A'], c=row['z'], norm=norm, cmap=colormap,edgecolor='k', lw=0.8, s=80)

plt.xticks(timerange, rotation=90)
ax.xaxis.set_major_formatter(mdates.DateFormatter("%d/%m/%Y"))
plt.xlabel('Time', fontsize=11, color='k')

clb = fig.colorbar(ax)       
clb.ax.set_title('Value', y=-0.125, fontsize=11)
clb.ax.invert_yaxis()

fig.tight_layout()

this produces AttributeError: 'AxesSubplot' object has no attribute 'autoscale_None'

but if I specify my ax as the scatter plot so that I can get my colour-coding working, I then have trouble with the axis formatter. Writing instead ax = plt.scatter generates AttributeError: 'PathCollection' object has no attribute 'xaxis' .

How can I have both the colorbar AND formatted axis ticks?

Answer 1

Don't call the scatter ax . (This overwrites the existinge axes ax .)
The colorbar expects as first argument a ScalarMappable (as eg the scatter). Since the scatters are all normalized, you can use it from the loop,

norm = plt.Normalize(...)
for bla in blubb:
    scatter = plt.scatter(..., norm=norm) 

Then,

clb = fig.colorbar(scatter)

The rest should stay the same.

Answer 2

The basic idea is that you need to add an extra axis for the colorbar.

It's hard to know if this is an exact match, as you haven't provided a working example with data. But this may at least serve as a template.

First, some example data:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import matplotlib.cm as cm
import matplotlib.dates as mdates
from mpl_toolkits.axes_grid1 import make_axes_locatable

vmin = 0
vmax = 1000
timerange = pd.date_range(start='2015-01-01', end='2016-01-01', freq='30D')
N = len(timerange)

data = np.random.randint(vmin, vmax, size=N)
# z contains the colorbar values for each point
cmap = plt.get_cmap('Reds')
z = [cmap((x-vmin)/(vmax-vmin))[:3] for x in data]
df = pd.DataFrame({"value":data, "datetime":timerange, "z":z})

Now plot:

fig = plt.figure(figsize=(6.,5))
ax = fig.add_subplot(111)

plt.scatter(x=df.datetime.values, y=df.value.values, c=df.z)

ax.set_xticklabels(timerange, rotation=90)
ax.xaxis.set_major_formatter(mdates.DateFormatter("%d/%m/%Y"))
ax.set_xlabel('Time')

Now add colorbar:

norm = mcolors.Normalize(vmin=vmin,vmax=vmax)
m = cm.ScalarMappable(cmap='Reds', norm=norm)
m.set_array([(x-vmin)/(vmax-vmin) for x in df.value.values])

divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)
clb = plt.colorbar(m, cax=cax)