pandas plot自动为类别分配颜色

Question

我想做一个条形图我的数据框，这样分类列 ('ad') 定义了我的条形图的颜色。 这是我的数据：

"date","shown","clicked","converted","avg_cost_per_click","total_revenue","ad"
2015-10-01,65877,2339,43,0.9,641.62,"ad_group_1"
2015-10-02,65100,2498,38,0.94,756.37,"ad_group_1"
2015-10-03,70658,2313,49,0.86,970.9,"ad_group_2"
2015-10-04,69809,2833,51,1.01,907.39,"ad_group_2"
2015-10-05,68186,2696,41,1,879.45,"ad_group_3"
2015-10-06,66864,2617,46,0.98,746.48,"ad_group_3"
2015-10-07,68227,2390,42,0.94,462.33,"ad_group_4"
2015-10-08,68520,2909,46,1.07,441.28,"ad_group_4"
2015-10-09,67250,2385,49,0.88,602.14,"ad_group_5"

我有多达 40 个广告组。

我的代码：

columns=['date','converted','clicked','ad']
df2=pd.DataFrame(df,columns=columns)
df2.set_index(df2.date,inplace=True)

这里我只取 1 列（“点击”）来绘制“日期”

plt.figure()
df2.loc[:,['clicked','ad']].plot(kind='bar')
plt.show()
#df2.loc[:,['clicked','ad']].plot(kind='bar',colormap='ad')

colormap 的代码行不起作用，因为我没有颜色列表。 我已经看到用户手动创建类别到颜色的颜色映射的答案。我有 40 多个类别，无法手动完成。

有什么方法可以将“ad”列的不同值设置为自动分配颜色？

寻找这样的东西：

Answer 1

您可以获得ad 列中存在的每个组的不同颜色，如图所示：

df.set_index(['date'],inplace=True) 

# Empty list to append later
grouped_list = []
label_list = []

# Iterating through each group key grouped by ad
for label, key in df.groupby(['ad'])['clicked']:
    grouped_list.append(key)
    label_list.append(label)

# Concatenating the grouped list column-wise and filling Nans with 0's 
df_grouped_bar = pd.concat(grouped_list, axis=1).fillna(0)

# Renaming columns created to take on new names
df_grouped_bar.columns = label_list

# Bar plot with a chosen colormap
ax = df_grouped_bar.plot(kind='bar', stacked=True, figsize=(6,6), 
                         width=0.2, ylim=(0,5000), cmap=plt.cm.rainbow)
# Figure Aesthetics
ax.set_xticklabels(df_grouped_bar.index.format())
plt.ylabel('clicked')
plt.tight_layout()
plt.show()