使用matplotlib的水平条形图

Question

我有数据为avito_trend.csv的文件，我想打印

import pandas as pd
import itertools

df = pd.read_csv("avito_trend.csv", parse_dates=[2])

def f(df):
    dfs = []
    for x in [list(x) for x in itertools.combinations(df['address'].unique(), 2)]:    
        c1 = df.loc[df['address'].isin([x[0]]), 'ID']
        c2 = df.loc[df['address'].isin([x[1]]), 'ID']
        c = pd.Series(list(set(c1).intersection(set(c2))))
        dfs.append(pd.DataFrame({'a':len(c), 'b':' and '.join(x)}, index=[0]))
    return pd.concat(dfs)

result = df.groupby([df['used_at'].dt.year]).apply(f).reset_index(drop=True, level=1).reset_index()
print result


    used_at     a                                    b
0      2014  1364                   avito.ru and e1.ru
1      2014  1716                 avito.ru and drom.ru
2      2014  1602                 avito.ru and auto.ru
3      2014   299           avito.ru and avtomarket.ru
4      2014   579                   avito.ru and am.ru
5      2014   602             avito.ru and irr.ru/cars
6      2014   424       avito.ru and cars.mail.ru/sale
7      2014   634                    e1.ru and drom.ru
8      2014   475                    e1.ru and auto.ru
9      2014   139              e1.ru and avtomarket.ru
10     2014   224                      e1.ru and am.ru
11     2014   235                e1.ru and irr.ru/cars
12     2014   154          e1.ru and cars.mail.ru/sale
13     2014   874                  drom.ru and auto.ru
14     2014   247            drom.ru and avtomarket.ru
15     2014   394                    drom.ru and am.ru
16     2014   423              drom.ru and irr.ru/cars
17     2014   292        drom.ru and cars.mail.ru/sale
18     2014   243            auto.ru and avtomarket.ru
19     2014   408                    auto.ru and am.ru
20     2014   409              auto.ru and irr.ru/cars
21     2014   330        auto.ru and cars.mail.ru/sale
22     2014   133              avtomarket.ru and am.ru
23     2014   139        avtomarket.ru and irr.ru/cars
24     2014   105  avtomarket.ru and cars.mail.ru/sale
25     2014   223                am.ru and irr.ru/cars
26     2014   166          am.ru and cars.mail.ru/sale
27     2014   197    irr.ru/cars and cars.mail.ru/sale
28     2015  1153                   avito.ru and e1.ru
29     2015  1473                 avito.ru and auto.ru
30     2015  1491                 avito.ru and drom.ru
31     2015   403             avito.ru and irr.ru/cars
32     2015   205           avito.ru and avtomarket.ru
33     2015   256       avito.ru and cars.mail.ru/sale
34     2015   262                   avito.ru and am.ru
35     2015   451                    e1.ru and auto.ru
36     2015   539                    e1.ru and drom.ru
37     2015   148                e1.ru and irr.ru/cars
38     2015   105              e1.ru and avtomarket.ru
39     2015   105          e1.ru and cars.mail.ru/sale
40     2015    99                      e1.ru and am.ru
41     2015   799                  auto.ru and drom.ru
42     2015   288              auto.ru and irr.ru/cars
43     2015   162            auto.ru and avtomarket.ru
44     2015   195        auto.ru and cars.mail.ru/sale
45     2015   224                    auto.ru and am.ru
46     2015   277              drom.ru and irr.ru/cars
47     2015   175            drom.ru and avtomarket.ru
48     2015   189        drom.ru and cars.mail.ru/sale
49     2015   187                    drom.ru and am.ru
50     2015    73        irr.ru/cars and avtomarket.ru
51     2015    94    irr.ru/cars and cars.mail.ru/sale
52     2015   102                irr.ru/cars and am.ru
53     2015    48  avtomarket.ru and cars.mail.ru/sale
54     2015    72              avtomarket.ru and am.ru
55     2015    73          cars.mail.ru/sale and am.ru

我使用ax = result.plot(width=0.5, kind='barh', stacked=True) 我想得到类似this graph的smth。 我该怎么做？ 我需要的是2014年和2015年的意义，为另一对网站。 相反，我需要第a列的平均值

Answer 1

正如@user308827已经说过的，我也会使用seaborn来实现这一点，但我会做得有点不同：

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

pd.set_option('display.expand_frame_repr', False)

cols = ['ID', 'address', 'used_at']
df = pd.read_csv(r'D:\data\gDrive\data\.stack.overflow\data\avito_trend.csv.gz',
                 parse_dates=['used_at'], usecols=cols)

df.sort_values(['ID','used_at','address'], inplace=True)

df['prev_address'] = df['address'].shift()
df['time_diff'] = df['used_at'] - df['used_at'].shift()

df = df[df['address'] != df['prev_address']]
df = df[df['time_diff'] <= pd.Timedelta('10min')]


tmp = df[['ID','address','prev_address']] \
      .groupby(['address','prev_address', df.used_at.dt.year]) \
      .count() \
      .reset_index()

# remove `df` from memory
del df

tmp['visit_from'] = tmp['prev_address'] + ' -> ' + tmp['address']

# keep only 'interesting' columns
tmp = tmp[['visit_from','used_at','ID']]
tmp.columns = ['visit_from','year','visits']

# save temporary groupped CSV file
#fn = r'D:\data\gDrive\data\.stack.overflow\data\avito_grp.csv'
#tmp.to_csv(fn, index=False)

# show all
#df = tmp

# show only those sites with visits >= 100 (within both years)
df = tmp[tmp.groupby('visit_from')['visits'].transform('sum') >= 100].reset_index()

# prepare sorted index
idx = df.groupby('visit_from')['visits'].transform('sum').sort_values(ascending=False).index

# 'apply' index
df = df.reindex(idx)

# add 'total' column (sum of visits for all years)
#df['total'] = df.groupby('visit_from')['visits'].transform('sum')

################################################
#
# SeaBorn plotting
#
sns.set(style="darkgrid")
sns.set_color_codes("pastel")

f, ax = plt.subplots(figsize=(16, 12))
ax = sns.barplot(x='visits', y='visit_from', hue='year', data=df, saturation=0.8)
plt.xlabel('Visits')

# add annotations
[ax.text(p.get_width() + 3, p.get_y() + p.get_height(),
         int(p.get_width()), fontsize=8)
 for p in ax.patches]


plt.show()

PS您感兴趣的部分以SeaBorn plotting注释开头

Answer 2

尝试seaborn

来自https://stanford.edu/~mwaskom/software/seaborn/generated/seaborn.barplot.html

import seaborn as sns
sns.set_style("whitegrid")
tips = sns.load_dataset("tips")
ax = sns.barplot(x="day", y="total_bill", data=tips)

对于堆积条形图： https://gist.github.com/randyzwitch/b71d47e0d380a1a6bef9