第07章 数据清洗和准备
7.1 处理缺失数据
pandas使用浮点值NaN（Not a Number）表示缺失数据。
Python内置的None值在对象数组中也可以作为NA.
列出了一些关于缺失值处理的函数。
import pandas as pd
import numpy as np
string_data = pd.Series(['aardvark', 'artichoke', np.nan, 'avocado'])
string_data
0 aardvark
1 artichoke
2 NaN
3 avocado
dtype: object
string_data.isnull()
0 False
1 False
2 True
3 False
dtype: bool
string_data[0] = None
string_data.isnull()
0 True
1 False
2 True
3 False
dtype: bool
滤除缺失数据
from numpy import nan as NA
data = pd.Series([1, NA, 3.5, NA, 7])
data.dropna()
0 1.0
2 3.5
4 7.0
dtype: float64
#等价于:
data[data.notnull()]
0 1.0
2 3.5
4 7.0
dtype: float64
#对于DataFrame对象，dropna默认丢弃任何含有缺失值的行.
data = pd.DataFrame([[1., 6.5, 3.], [1., NA, NA],
[NA, NA, NA], [NA, 6.5, 3.]])
cleaned = data.dropna()
print(data)
print('\n')
print('\n')
print(cleaned)
0 1 2
0 1.0 6.5 3.0
1 1.0 NaN NaN
2 NaN NaN NaN
3 NaN 6.5 3.0
0 1 2
0 1.0 6.5 3.0
#传入how='all'将只丢弃全为NA的那些行：
print(data.dropna(how='all'))
0 1 2
0 1.0 6.5 3.0
1 1.0 NaN NaN
3 NaN 6.5 3.0
#用这种方式丢弃列，只需传入axis=1即可：
data[4] = NA
print(data)
print('\n')
print(data.dropna(axis=1, how='all'))
0 1 2 4
0 1.0 6.5 3.0 NaN
1 1.0 NaN NaN NaN
2 NaN NaN NaN NaN
3 NaN 6.5 3.0 NaN
0 1 2
0 1.0 6.5 3.0
1 1.0 NaN NaN
2 NaN NaN NaN
3 NaN 6.5 3.0
#只想留下1部分观测数据，可以用thresh参数实现此目的
df = pd.DataFrame(np.random.randn(7, 3))
df.iloc[:4, 1] = NA
df.iloc[:2, 2] = NA
print(df)
print('\n')
print(df.dropna())
print('\n')
print(df.dropna(thresh=2))
0 1 2
0 0.875651 NaN NaN
1 0.261047 NaN NaN
2 -0.677917 NaN 0.360858
3 -1.035602 NaN 0.097730
4 0.127693 -1.006822 0.404568
5 -0.124359 0.444798 -0.028498
6 -0.779734 -0.012785 1.745280
0 1 2
4 0.127693 -1.006822 0.404568
5 -0.124359 0.444798 -0.028498
6 -0.779734 -0.012785 1.745280
0 1 2
2 -0.677917 NaN 0.360858
3 -1.035602 NaN 0.097730
4 0.127693 -1.006822 0.404568
5 -0.124359 0.444798 -0.028498
6 -0.779734 -0.012785 1.745280
填充缺失数据
通过1个常数调用fillna就会将缺失值替换为那个常数值。
print(df.fillna(0))
0 1 2
0 0.875651 0.000000 0.000000
1 0.261047 0.000000 0.000000
2 -0.677917 0.000000 0.360858
3 -1.035602 0.000000 0.097730
4 0.127693 -1.006822 0.404568
5 -0.124359 0.444798 -0.028498
6 -0.779734 -0.012785 1.745280
#可以实现对不同列填充不同的值
print(df.fillna({1: 0.5, 2: 0}))
#fillna默认会返回新对象，但也可以对现有对象进行就地修改：
0 1 2
0 0.875651 0.500000 0.000000
1 0.261047 0.500000 0.000000
2 -0.677917 0.500000 0.360858
3 -1.035602 0.500000 0.097730
4 0.127693 -1.006822 0.404568
5 -0.124359 0.444798 -0.028498
6 -0.779734 -0.012785 1.745280
#reindexing有效的插值方法也可以用于fillna。
df = pd.DataFrame(np.random.randn(7, 3))
df.iloc[2:, 1] = NA
df.iloc[4:, 2] = NA
print(df)
print('\n')
print(df.fillna(method='ffill'))
print('\n')
print(df.fillna(method='ffill', limit=2))
0 1 2
0 0.210105 0.442817 -0.468990
1 -1.027097 0.892226 -0.452749
2 0.662849 NaN -0.476516
3 0.556252 NaN 0.208664
4 -0.021977 NaN NaN
5 -1.325938 NaN NaN
6 -0.520716 NaN NaN
0 1 2
0 0.210105 0.442817 -0.468990
1 -1.027097 0.892226 -0.452749
2 0.662849 0.892226 -0.476516
3 0.556252 0.892226 0.208664
4 -0.021977 0.892226 0.208664
5 -1.325938 0.892226 0.208664
6 -0.520716 0.892226 0.208664
0 1 2
0 0.210105 0.442817 -0.468990
1 -1.027097 0.892226 -0.452749
2 0.662849 0.892226 -0.476516
3 0.556252 0.892226 0.208664
4 -0.021977 NaN 0.208664
5 -1.325938 NaN 0.208664
6 -0.520716 NaN NaN
7.2 数据转换
移除重复数据
data = pd.DataFrame({'k1': ['one', 'two'] * 3 + ['two'],
'k2': [1, 1, 2, 3, 3, 4, 4]})
print(data)
k1 k2
0 one 1
1 two 1
2 one 2
3 two 3
4 one 3
5 two 4
6 two 4
#DataFrame的duplicated方法返回1个布尔型Series，表示各⾏是否是重复行（前面出现过的行）
print(data.duplicated())
#还有1个与此相关的drop_duplicates方法，它会返回1个DataFrame，重复的数组会标为False：
print('\n')
print( data.drop_duplicates())
0 False
1 False
2 False
3 False
4 False
5 False
6 True
dtype: bool
k1 k2
0 one 1
1 two 1
2 one 2
3 two 3
4 one 3
5 two 4
#假设我们还有1列值，且只希望根据k1列过滤重复项：
data['v1'] = range(7)
print(data.drop_duplicates(['k1']))
#duplicated和drop_duplicates默认保留的是第1个出现的值组合。传入keep='last'则保留最后1个
print('\n')
print(data.drop_duplicates(['k1', 'k2'], keep='last'))
k1 k2 v1
0 one 1 0
1 two 1 1
k1 k2 v1
0 one 1 0
1 two 1 1
2 one 2 2
3 two 3 3
4 one 3 4
6 two 4 6
利用函数或映射进行数据转换
data = pd.DataFrame({'food': ['bacon', 'pulled pork', 'bacon',
'Pastrami', 'corned beef', 'Bacon',
'pastrami', 'honey ham', 'nova lox'],
'ounces': [4, 3, 12, 6, 7.5, 8, 3, 5, 6]})
print(data)
food ounces
0 bacon 4.0
1 pulled pork 3.0
2 bacon 12.0
3 Pastrami 6.0
4 corned beef 7.5
5 Bacon 8.0
6 pastrami 3.0
7 honey ham 5.0
8 nova lox 6.0
#你想要添加1列表示该肉类食物来源的动物类型。我们先编写1个不同⾁类到动物的映射
meat_to_animal = {
'bacon': 'pig',
'pulled pork': 'pig',
'pastrami': 'cow',
'corned beef': 'cow',
'honey ham': 'pig',
'nova lox': 'salmon'
}
#使用Series的str.lower方法，将各个值转换为小写
lowercased = data['food'].str.lower()
print(lowercased)
data['animal'] = lowercased.map(meat_to_animal)
print('\n')
print(data)
0 bacon
1 pulled pork
2 bacon
3 pastrami
4 corned beef
5 bacon
6 pastrami
7 honey ham
8 nova lox
Name: food, dtype: object
food ounces animal
0 bacon 4.0 pig
1 pulled pork 3.0 pig
2 bacon 12.0 pig
3 Pastrami 6.0 cow
4 corned beef 7.5 cow
5 Bacon 8.0 pig
6 pastrami 3.0 cow
7 honey ham 5.0 pig
8 nova lox 6.0 salmon
替换值
data = pd.Series([1., -999., 2., -999., -1000., 3.])
print(data)
data.replace(-999, np.nan)
0 1.0
1 -999.0
2 2.0
3 -999.0
4 -1000.0
5 3.0
dtype: float64
0 1.0
1 NaN
2 2.0
3 NaN
4 -1000.0
5 3.0
dtype: float64
#一次性替换多个值
print(data.replace([-999, -1000], np.nan))#一换多
print('\n')
print(data.replace([-999, -1000], [np.nan, 0]))#一换一,列表形式
print('\n')
print( data.replace({-999: np.nan, -1000: 0}))#一换一,字典形式
0 1.0
1 NaN
2 2.0
3 NaN
4 NaN
5 3.0
dtype: float64
0 1.0
1 NaN
2 2.0
3 NaN
4 0.0
5 3.0
dtype: float64
0 1.0
1 NaN
2 2.0
3 NaN
4 0.0
5 3.0
dtype: float64
重命名轴索引
data = pd.DataFrame(np.arange(12).reshape((3, 4)),
index=['Ohio', 'Colorado', 'New York'],
columns=['one', 'two', 'three', 'four'])
transform = lambda x: x[:4].upper()
data.index.map(transform)
Index(['OHIO', 'COLO', 'NEW '], dtype='object')
data.index = data.index.map(transform)
print(data)
one two three four
OHIO 0 1 2 3
COLO 4 5 6 7
NEW 8 9 10 11
print(data.rename(index=str.title, columns=str.upper))
ONE TWO THREE FOUR
Ohio 0 1 2 3
Colo 4 5 6 7
New 8 9 10 11
#rename可以结合字典型对象实现对部分轴标签的更新
data.rename(index={'OHIO': 'INDIANA'},
columns={'three': 'peekaboo'})
| one | two | peekaboo | four | |
|---|---|---|---|---|
| INDIANA | 0 | 1 | 2 | 3 |
| COLO | 4 | 5 | 6 | 7 |
| NEW | 8 | 9 | 10 | 11 |
离散化和面元划分
ages = [20, 22, 25, 27, 21, 23, 37, 31, 61, 45, 41, 32]
#接下来将这些数据划分为“18到25”、“26到35”、“35到60”以及“60以上”几个面元。要实现该功能，你需
#要使用pandas的cut函数
bins = [18, 25, 35, 60, 100]
cats = pd.cut(ages, bins)
print(cats)
[(18, 25], (18, 25], (18, 25], (25, 35], (18, 25], ..., (25, 35], (60, 100], (35, 60], (35, 60], (25, 35]]
Length: 12
Categories (4, interval[int64]): [(18, 25] < (25, 35] < (35, 60] < (60, 100]]
cats.codes
array([0, 0, 0, 1, 0, 0, 2, 1, 3, 2, 2, 1], dtype=int8)
cats.categories
IntervalIndex([(18, 25], (25, 35], (35, 60], (60, 100]],
closed='right',
dtype='interval[int64]')
print(pd.value_counts(cats))
#pd.value_counts(cats)是pandas.cut结果的面元计数
(18, 25] 5
(35, 60] 3
(25, 35] 3
(60, 100] 1
dtype: int64
#跟“区间”的数学符号1样，圆括号表示开端，而方括号则表示闭端（包括）。哪边是闭端可以通过
#right=False进行修改
pd.cut(ages, [18, 26, 36, 61, 100], right=False)
[[18, 26), [18, 26), [18, 26), [26, 36), [18, 26), ..., [26, 36), [61, 100), [36, 61), [36, 61), [26, 36)]
Length: 12
Categories (4, interval[int64]): [[18, 26) < [26, 36) < [36, 61) < [61, 100)]
group_names = ['Youth', 'YoungAdult', 'MiddleAged', 'Senior']
pd.cut(ages, bins, labels=group_names)
[Youth, Youth, Youth, YoungAdult, Youth, ..., YoungAdult, Senior, MiddleAged, MiddleAged, YoungAdult]
Length: 12
Categories (4, object): [Youth < YoungAdult < MiddleAged < Senior]
data = np.random.rand(20)
#将1些均匀分布的数据分成四组
pd.cut(data, 4, precision=2)
#选项precision=2，限定小数只有两位
[(0.5, 0.74], (0.0012, 0.25], (0.25, 0.5], (0.5, 0.74], (0.25, 0.5], ..., (0.74, 0.99], (0.74, 0.99], (0.25, 0.5], (0.74, 0.99], (0.25, 0.5]]
Length: 20
Categories (4, interval[float64]): [(0.0012, 0.25] < (0.25, 0.5] < (0.5, 0.74] < (0.74, 0.99]]
data = np.random.randn(1000)
cats = pd.qcut(data, 4)
print(cats)
print('\n')
print(pd.value_counts(cats))
[(-0.737, -0.0221], (0.744, 2.901], (0.744, 2.901], (0.744, 2.901], (0.744, 2.901], ..., (-0.737, -0.0221], (0.744, 2.901], (0.744, 2.901], (0.744, 2.901], (0.744, 2.901]]
Length: 1000
Categories (4, interval[float64]): [(-3.961, -0.737] < (-0.737, -0.0221] < (-0.0221, 0.744] < (0.744, 2.901]]
(0.744, 2.901] 250
(-0.0221, 0.744] 250
(-0.737, -0.0221] 250
(-3.961, -0.737] 250
dtype: int64
检测和过滤异常值
data = pd.DataFrame(np.random.randn(1000, 4))
print(data.describe())
0 1 2 3
count 1000.000000 1000.000000 1000.000000 1000.000000
mean 0.000800 -0.032331 0.023345 0.048245
std 1.003987 0.996681 1.007219 1.010186
min -3.091473 -3.506493 -3.055961 -3.951916
25% -0.666621 -0.701949 -0.652654 -0.610015
50% -0.022072 -0.045974 0.010023 0.061506
75% 0.660415 0.638646 0.696614 0.728876
max 2.867995 2.941726 3.496384 2.806226
col = data[2]
col[np.abs(col) > 3]
231 3.058974
265 3.058203
453 3.496384
723 3.127166
734 3.364669
826 -3.055961
Name: 2, dtype: float64
#要选出全部含有“超过3或－3的值”的行，你可以在布尔型DataFrame中使用any方法：
print(data[(np.abs(data) > 3).any(1)])
0 1 2 3
160 -0.009122 1.304577 -0.562316 -3.211438
231 0.326291 -0.965339 3.058974 -0.249949
265 0.293463 0.026279 3.058203 -0.690907
298 0.172609 -3.506493 1.028811 -1.075643
423 -0.316089 -1.273325 -1.538425 -3.951916
453 -0.654466 -0.017330 3.496384 -0.315267
616 -3.091473 -1.450217 1.190238 -1.281410
723 -0.836537 -0.184380 3.127166 0.515251
734 1.096555 0.346592 3.364669 1.115433
826 0.278241 2.690008 -3.055961 -0.471325
901 -0.399469 -3.126419 -1.882590 -1.159301
#将值限制在-3到3之间的区间里
data[np.abs(data) > 3] = np.sign(data) * 3
print(data.describe())
0 1 2 3
count 1000.000000 1000.000000 1000.000000 1000.000000
mean 0.000891 -0.031698 0.022295 0.049408
std 1.003709 0.994655 1.003572 1.006192
min -3.000000 -3.000000 -3.000000 -3.000000
25% -0.666621 -0.701949 -0.652654 -0.610015
50% -0.022072 -0.045974 0.010023 0.061506
75% 0.660415 0.638646 0.696614 0.728876
max 2.867995 2.941726 3.000000 2.806226
#np.sign(data)可以生成1和-1：
print(np.sign(data).head())
0 1 2 3
0 1.0 1.0 -1.0 -1.0
1 -1.0 1.0 1.0 -1.0
2 1.0 -1.0 1.0 1.0
3 -1.0 1.0 1.0 1.0
4 -1.0 1.0 -1.0 1.0
排列和随机采样
df = pd.DataFrame(np.arange(5 * 4).reshape((5, 4)))
sampler = np.random.permutation(5)
print(sampler)
print('\n')
print(df)
print('\n')
print( df.take(sampler))
print('\n')
print(df.sample(n=3))
[1 4 3 2 0]
0 1 2 3
0 0 1 2 3
1 4 5 6 7
2 8 9 10 11
3 12 13 14 15
4 16 17 18 19
0 1 2 3
1 4 5 6 7
4 16 17 18 19
3 12 13 14 15
2 8 9 10 11
0 0 1 2 3
0 1 2 3
3 12 13 14 15
0 0 1 2 3
1 4 5 6 7
计算指标/哑变量
df = pd.DataFrame({'key': ['b', 'b', 'a', 'c', 'a', 'b'],
'data1': range(6)})
print( pd.get_dummies(df['key']))
a b c
0 0 1 0
1 0 1 0
2 1 0 0
3 0 0 1
4 1 0 0
5 0 1 0
dummies = pd.get_dummies(df['key'], prefix='key')
df_with_dummy = df[['data1']].join(dummies)
print(df_with_dummy)
data1 key_a key_b key_c
0 0 0 1 0
1 1 0 1 0
2 2 1 0 0
3 3 0 0 1
4 4 1 0 0
5 5 0 1 0
mnames = ['movie_id', 'title', 'genres']
movies = pd.read_table('datasets/movielens/movies.dat', sep='::',
header=None, names=mnames)
print(movies[:10])
movie_id title genres
0 1 Toy Story (1995) Animation|Children's|Comedy
1 2 Jumanji (1995) Adventure|Children's|Fantasy
2 3 Grumpier Old Men (1995) Comedy|Romance
3 4 Waiting to Exhale (1995) Comedy|Drama
4 5 Father of the Bride Part II (1995) Comedy
5 6 Heat (1995) Action|Crime|Thriller
6 7 Sabrina (1995) Comedy|Romance
7 8 Tom and Huck (1995) Adventure|Children's
8 9 Sudden Death (1995) Action
9 10 GoldenEye (1995) Action|Adventure|Thriller
E:\anaconda\lib\site-packages\ipykernel_launcher.py:3: ParserWarning: Falling back to the 'python' engine because the 'c' engine does not support regex separators (separators > 1 char and different from '\s+' are interpreted as regex); you can avoid this warning by specifying engine='python'.
This is separate from the ipykernel package so we can avoid doing imports until
#从数据集中抽取出不同的genre值
all_genres = []
for x in movies.genres:
all_genres.extend(x.split('|'))
genres = pd.unique(all_genres)
print(genres)
['Animation' "Children's" 'Comedy' 'Adventure' 'Fantasy' 'Romance' 'Drama'
'Action' 'Crime' 'Thriller' 'Horror' 'Sci-Fi' 'Documentary' 'War'
'Musical' 'Mystery' 'Film-Noir' 'Western']
zero_matrix = np.zeros((len(movies), len(genres)))
dummies = pd.DataFrame(zero_matrix, columns=genres)
#使用dummies.columns来计算每个类型的列索引
gen = movies.genres[0]
print(gen.split('|'))
dummies.columns.get_indexer(gen.split('|'))
['Animation', "Children's", 'Comedy']
array([0, 1, 2], dtype=int32)
for i, gen in enumerate(movies.genres):
indices = dummies.columns.get_indexer(gen.split('|'))
dummies.iloc[i, indices] = 1
movies_windic = movies.join(dummies.add_prefix('Genre_'))
movies_windic.iloc[0]
movie_id 1
title Toy Story (1995)
genres Animation|Children's|Comedy
Genre_Animation 1
Genre_Children's 1
Genre_Comedy 1
Genre_Adventure 0
Genre_Fantasy 0
Genre_Romance 0
Genre_Drama 0
Genre_Action 0
Genre_Crime 0
Genre_Thriller 0
Genre_Horror 0
Genre_Sci-Fi 0
Genre_Documentary 0
Genre_War 0
Genre_Musical 0
Genre_Mystery 0
Genre_Film-Noir 0
Genre_Western 0
Name: 0, dtype: object
np.random.seed(12345)
values = np.random.rand(10)
bins = [0, 0.2, 0.4, 0.6, 0.8, 1]
print( pd.get_dummies(pd.cut(values, bins)))
(0.0, 0.2] (0.2, 0.4] (0.4, 0.6] (0.6, 0.8] (0.8, 1.0]
0 0 0 0 0 1
1 0 1 0 0 0
2 1 0 0 0 0
3 0 1 0 0 0
4 0 0 1 0 0
5 0 0 1 0 0
6 0 0 0 0 1
7 0 0 0 1 0
8 0 0 0 1 0
9 0 0 0 1 0
7.3 字符串操作
字符串对象⽅法
#split分割数据
val = 'a,b, guido'
val.split(',')
['a', 'b', ' guido']
#split常常与strip1起使用，以去除空白符（包括换行符）
pieces = [x.strip() for x in val.split(',')]
print(pieces)
['a', 'b', 'guido']
first, second, third = pieces
first + '::' + second + '::' + third
'a::b::guido'
正则表达式
描述1个或多个空白符的regex是\s+
import re
text = "foo bar\t baz \tqux"
re.split('\s+', text)
['foo', 'bar', 'baz', 'qux']
#findall返回的是字符串中所有的匹配项，而search则只返回第1个匹配项。
text = """Dave dave@google.com
Steve steve@gmail.com
Rob rob@gmail.com
Ryan ryan@yahoo.com
"""
pattern = r'[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}'
# re.IGNORECASE makes the regex case-insensitive
regex = re.compile(pattern, flags=re.IGNORECASE)
regex.findall(text)
['dave@google.com', 'steve@gmail.com', 'rob@gmail.com', 'ryan@yahoo.com']
m = regex.search(text)
print(m)
text[m.start():m.end()]
<re.Match object; span=(5, 20), match='dave@google.com'>
'dave@google.com'
pandas的矢量化字符串函数
data = {'Dave': 'dave@google.com', 'Steve': 'steve@gmail.com','Rob': 'rob@gmail.com', 'Wes': np.nan}
data = pd.Series(data)
print(data)
print('\n')
print(data.isnull)
Dave dave@google.com
Steve steve@gmail.com
Rob rob@gmail.com
Wes NaN
dtype: object
<bound method Series.isnull of Dave dave@google.com
Steve steve@gmail.com
Rob rob@gmail.com
Wes NaN
dtype: object>
pattern = r'[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}'
print(data.str.contains('gmail'))
print('\n')
print(pattern)
print('\n')
print(data.str.findall(pattern, flags=re.IGNORECASE))
Dave False
Steve True
Rob True
Wes NaN
dtype: object
[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}
Dave [dave@google.com]
Steve [steve@gmail.com]
Rob [rob@gmail.com]
Wes NaN
dtype: object
3255
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
