Pandas bar plot:添加标记以区分0和NaN

时间:2015-07-22 06:50:33

标签: python pandas matplotlib plot nan

我在Python中有以下pandas DataFrame:

size

我想使用df = pd.DataFrame([10,0,np.nan,8],index=[1947,1948,1949,1950], columns=['values']) df values 1947 10 1948 0 1949 NaN 1950 8 做一个条形图。

如何添加一种标记来说明df.plot(kind='bar')0之间的区别(并将其添加到图例中)?

通用结果如下所示: Bar plot of the values from the dataframe, showing 1947 as 10, 1948 as 0, 1949 as 0 and 1950 as 8

编辑:嗯,最好的是: enter image description here

我尝试使用'scatter'选项结合firelynx的第一个解决方案,但仍然遇到一些错误......

2 个答案:

答案 0 :(得分:1)

如果您没有任何具体要求,可以很容易地想象出nans,这是以下几种方法之一:

df['isnan'] = pd.isnull(df['values'])
df.plot(kind='bar')

enter image description here

这是另一种方式:

df['values'] = df['values'].fillna(-1)
df.plot(kind='bar')

enter image description here

与大熊猫一样,一旦你想要一种完全一种方式的东西,它就会复杂得多。

import matplotlib.pyplot as plt
import matplotlib.ticker as plticker

markers = df[df['isnan']]
fig, ax1 = plt.subplots()
ax1.bar(df.index, df['values'], 0.4, align='center')       
ax1.plot(markers.index, markers, 'ro')
loc = plticker.MultipleLocator(base=1.0)
ax1.xaxis.set_major_locator(loc)
ax1.xaxis.set_ticklabels(["","1947", "1948", "1949", "1950"])
plt.show()

enter image description here

答案 1 :(得分:1)

enter image description here

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches

df = pd.DataFrame([10,0,np.nan,8],index=[1947,1948,1949,1950], columns=['values'])
ax = df.plot(kind='bar', color='gray', alpha=0.5)
ax.get_figure().set_facecolor('white')

nan_idx = np.where(df['values'].isnull())[0]
plt.axvspan(nan_idx-0.25, nan_idx+0.25, facecolor='white', alpha=0.5, hatch='X')

nan_legend = mpatches.Patch(facecolor='white', edgecolor='gray', hatch='X', label='nan Value')
ordinary_legend = mpatches.Patch(color='gray', label='ordinary Value')
plt.legend(handles=[nan_legend, ordinary_legend])

plt.show()