如何正确地将Seaborn / matplotlib中的x轴与时间轴隔开？

Question

I have a pandas dataframe with the time-axis set as the index. When plotting the following code

%matplotlib inline
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns; sns.set();
ids = [1,2,3,4]
dates = pd.date_range('20150704', '20160331')
vals = np.random.randn(len(ids)*len(dates))
allids = np.tile(ids, len(dates))
alldates = np.tile(dates, len(ids))
df = (pd.DataFrame(np.vstack((allids, vals)).T, index=alldates)
    .reset_index().rename(columns={'index':'datetime',0:'unique_id',1:'height'}))

time_group = '1w'
threshold = 0.50

fig, axes = plt.subplots(nrows=4, ncols=1, figsize=(12, 16), sharex=True)

for (i, temp), ax in zip(df.groupby('unique_id'), axes.ravel()):
    (temp.set_index('datetime').height
         .groupby(pd.TimeGrouper(time_group))
         .mean()
         .plot(ax=ax))

I get a plot with 4 plots. The spacing of the vertical lines on the x-axis looks like:

Notice how, for example, November is wider than December (so it isn't merely the number of days in each month). Apparently this is due to the number of data points in each month, not the number of days in each month. Any suggestions to make the vertical lines spaced with the start/end of the months properly in time? Is this a bug?

Answer 1

The 1W will group based on Sundays in each month and there are some months that we have more Sundays compare to others... That's why some months are wider. Below change should help...

    %matplotlib inline
    import matplotlib.pyplot as plt
    import numpy as np
    import pandas as pd
    import matplotlib.dates as mdates
    import seaborn as sns; sns.set();
    ids = [1,2,3,4]
    dates = pd.date_range('20150704', '20160331')
    vals = np.random.randn(len(ids)*len(dates))
    allids = np.tile(ids, len(dates))
    alldates = np.tile(dates, len(ids))
    df = (pd.DataFrame(np.vstack((allids, vals)).T, index=alldates)
        .reset_index().rename(columns={'index':'datetime',0:'unique_id',1:'height'}))
    time_group = '1w'
    threshold = 0.50

    fig, axes = plt.subplots(nrows=4, ncols=1, figsize=(12, 16), sharex=True)


    for (i, temp), ax in zip(df.groupby('unique_id'), axes.ravel()):
        b=(temp.set_index('datetime').height
             .groupby(pd.TimeGrouper(time_group))
             .mean())
        ax.plot(b.index,b)
        ax.xaxis.set_major_formatter(mdates.DateFormatter('%b'))
    plt.show()