Matplotlib应用xaxis和yaxis数字格式

Question

1. Why is my xaxis not formatted as a date? I expect it to because I am setting the dataframe index to a datetime index.
2. How do I make all subplots share the same xaxis? The reason I am currently using add_subplot instead of plt.subplots is because I couldn't get plt.subplots to work for what I want to do -- which is to make nrows and ncols dynamic parameters so I can output charts with any shape I want: (4,1) shape, (2,2), (1,4), etc.
3. How do I apply a specific number format to each yaxis? Below, I am attempting to lookup the plot string in d (dict) to return the format string and then apply that format to the yaxis formatting but it doesn't seem to be working.

---

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.ticker import FuncFormatter
plt.style.use('ggplot')

df = pd.DataFrame({'Clicks': {0: 334, 1: 554, 2: 433, 3: 5353, 4: 433},
                   'Impressions': {0: 3242, 1: 43345, 2: 3456, 3: 34543, 4: 3453},
                   'Day': {0: '12/1/2015', 1: '12/2/2015', 2: '12/3/2015', 3: '12/4/2015', 4: '12/5/2015'},
                   'Conv': {0: 23, 1: 23, 2: 45, 3: 56, 4: 45},
                   'Cost': {0: 414.16, 1: 686.96, 2: 536.91, 3: 637.72, 4: 536.91}},
                  columns=['Day', 'Impressions', 'Clicks', 'Cost', 'Conv'])

df['Day'] = pd.to_datetime(df['Day'])
df = df.set_index('Day').resample('d', how='sum')

window = 2
nrows = 2
ncols = 2

plots = ['Impressions', 'Clicks', 'Cost', 'Conv']
d = {'Impressions':'{:,.0f}', 'Clicks': '{:,.0f}', 'Cost':'${:,.2f}', 'Conv': '{:,.0f}'}

fig = plt.figure(figsize=(8,6))
for i, plot in enumerate(plots):
  ax = fig.add_subplot(nrows, ncols, i+1)
  ax.plot(df.index, df[plot])
  ma = pd.rolling_mean(df[plot], window)
  ax.plot(df.index, ma)
  mstd = pd.rolling_std(df[plot], window)
  ax.fill_between(df.index, ma - 2*mstd, ma + 2*mstd, color='b', alpha=0.1)
  ax.set_title(plot)
  ax.get_yaxis().set_major_formatter(FuncFormatter(lambda x, p: d[plot].format(x)))
  plt.tight_layout()

plt.show()

Here's the df :

            Impressions  Clicks    Cost  Conv
Day                                          
2015-12-01         3242     334  414.16    23
2015-12-02        43345     554  686.96    23
2015-12-03         3456     433  536.91    45
2015-12-04        34543    5353  637.72    56
2015-12-05         3453     433  536.91    45

Answer 1

Why is my xaxis not formatted as a date?

You need to set DateFormatter (or similar) as major_formatter - see code below.

How do I make all subplots share the same xaxis?

Add sharex=True parameter to you subplots call. You can use axes from .subplots() if you flatten them like it is shown in the code below.

How do I apply a specific number format to each yaxis?

Your FuncFormatter needs to return a formatted string from given tick_value and position like in the code below:

fig, axes = plt.subplots(2, 2, figsize=(8,6), sharex=True)

for ax, plot in zip(axes.flat, plots):
    ax.plot(df.index, df[plot])
    ma = pd.rolling_mean(df[plot], window)
    ax.plot(df.index, ma)
    mstd = pd.rolling_std(df[plot], window)
    ax.fill_between(df.index, ma - 2*mstd, ma + 2*mstd, color='b', alpha=0.1)
    ax.set_title(plot)
    ax.yaxis.set_major_formatter(FuncFormatter(lambda x, p: '{:.0f}'.format(x)))
    ax.xaxis.set_major_formatter(DateFormatter('%d-%H:%M')) # or '%d.%m.%y'

fig.autofmt_xdate()  # This will rotate the xticklabels by 30 degrees so that all dates are readable.
fig.tight_layout()  # no need to call this inside the loop.

This will produce a plot like this: