在Seaborn的Regplot中使用Datetimes

Question

我正在Jupyter / IPython工作，每天绘制一定数量的单词，但是在Seaborn中使用Regplot的日期时却遇到了麻烦。 Regplot本身显然是does not support regression against date data，虽然我想要完成的并不一定需要Regplot的解决方法 - 也许只是一种格式化x轴标签的方法。

一个最小的工作示例，使用简单的时间戳：

%matplotlib inline
import numpy as np
import pandas as pd
import matplotlib as mpl
import matplotlib.pyplot as plt 
import matplotlib.dates as dates
import seaborn as sns
import time
import datetime
import radar
sns.set(style="whitegrid", color_codes=True)

data = pd.DataFrame([])

for i in np.arange(1, 10):
    date =  radar.random_datetime(start='2016-05-20', stop='2016-05-25')
    data = data.append(pd.DataFrame({'Date': time.mktime(date.timetuple()), 'Words': i + 100}, index=[0]), ignore_index=True)

points = plt.scatter(x = data['Date'], y = data["Words"], c=data["Words"], s=75, cmap="BrBG")
plt.colorbar(points)
sns.regplot(x = data['Date'], y = data["Words"], data=data, scatter=False, color='r')

使用重叠的趋势线渲染散点图：

但是将日期作为日期时间：

points = plt.scatter(x = pd.to_datetime(data['Date'], unit='s').dt.to_pydatetime(), y = data["Words"], c=data["Words"], s=75, cmap="BrBG")
plt.colorbar(points)
sns.regplot(x = pd.to_datetime(data['Date'], unit='s').dt.to_pydatetime(), y = data["Words"], data=data, scatter=False, color='r')

返回时出现以下错误：

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-7-d6488afe3dcb> in <module>()
      1 points = plt.scatter(x = pd.to_datetime(data['Date'], unit='s').dt.to_pydatetime(), y = data["Words"], c=data["Words"], s=75, cmap="BrBG")
      2 plt.colorbar(points)
----> 3 sns.regplot(x = pd.to_datetime(data['Date'], unit='s').dt.to_pydatetime(), y = data["Words"], data=data, scatter=False, color='r')

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in regplot(x, y, data, x_estimator, x_bins, x_ci, scatter, fit_reg, ci, n_boot, units, order, logistic, lowess, robust, logx, x_partial, y_partial, truncate, dropna, x_jitter, y_jitter, label, color, marker, scatter_kws, line_kws, ax)
    777     scatter_kws["marker"] = marker
    778     line_kws = {} if line_kws is None else copy.copy(line_kws)
--> 779     plotter.plot(ax, scatter_kws, line_kws)
    780     return ax
    781 

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in plot(self, ax, scatter_kws, line_kws)
    330             self.scatterplot(ax, scatter_kws)
    331         if self.fit_reg:
--> 332             self.lineplot(ax, line_kws)
    333 
    334         # Label the axes

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in lineplot(self, ax, kws)
    375 
    376         # Fit the regression model
--> 377         grid, yhat, err_bands = self.fit_regression(ax)
    378 
    379         # Get set default aesthetics

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in fit_regression(self, ax, x_range, grid)
    207             yhat, yhat_boots = self.fit_logx(grid)
    208         else:
--> 209             yhat, yhat_boots = self.fit_fast(grid)
    210 
    211         # Compute the confidence interval at each grid point

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in fit_fast(self, grid)
    222         grid = np.c_[np.ones(len(grid)), grid]
    223         reg_func = lambda _x, _y: np.linalg.pinv(_x).dot(_y)
--> 224         yhat = grid.dot(reg_func(X, y))
    225         if self.ci is None:
    226             return yhat, None

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\seaborn\linearmodels.py in <lambda>(_x, _y)
    221         X, y = np.c_[np.ones(len(self.x)), self.x], self.y
    222         grid = np.c_[np.ones(len(grid)), grid]
--> 223         reg_func = lambda _x, _y: np.linalg.pinv(_x).dot(_y)
    224         yhat = grid.dot(reg_func(X, y))
    225         if self.ci is None:

C:\Python\WinPython-64bit-3.5.2.2Qt5\python-3.5.2.amd64\lib\site-packages\numpy\linalg\linalg.py in pinv(a, rcond)
   1614     a, wrap = _makearray(a)
   1615     _assertNoEmpty2d(a)
-> 1616     a = a.conjugate()
   1617     u, s, vt = svd(a, 0)
   1618     m = u.shape[0]

AttributeError: 'datetime.datetime' object has no attribute 'conjugate'

虽然散点图确实使用格式良好的日期时间进行渲染：

有没有办法在Regplot中使用日期时间，或者使用时间戳，但是将x轴上的标签格式化为日期？

Answer 1

您可以在xticks的位置获取时间戳的值，然后将其转换为所需的格式。

ax = plt.gca()
xticks = ax.get_xticks()
xticks_dates = [datetime.datetime.fromtimestamp(x).strftime('%Y-%m-%d %H:%M:%S') for x in xticks]
ax.set_xticklabels(xticks_dates)