初步探索

Python 中的探索性数据分析

Izzy Weber

Curriculum Manager, DataCamp

探索性数据分析

审查并清洗数据以…

提炼洞见
生成假设

书架上的许多书籍图

用 .head() 初看数据

books = pd.read_csv("books.csv")

books.head()

|                          name |                   author |  rating | year |       genre |
|-------------------------------|--------------------------|---------|------|-------------|
| 10-Day Green Smoothie Cleanse |                 JJ Smith |    4.73 | 2016 | Non Fiction |
|             11/22/63: A Novel |             Stephen King |    4.62 | 2011 |     Fiction |
|             12 Rules for Life |       Jordan B. Peterson |    4.69 | 2018 | Non Fiction |
|        1984 (Signet Classics) |            George Orwell |    4.73 | 2017 |     Fiction |
|          5,000 Awesome Facts  | National Geographic Kids |    4.81 | 2019 |   Childrens |

用 .info() 获取更多

books.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 350 entries, 0 to 349
Data columns (total 5 columns):
 #   Column  Non-Null Count  Dtype  
--   ------  --------------  -----  
 0   name    350 non-null    object 
 1   author  350 non-null    object 
 2   rating  350 non-null    float64
 3   year    350 non-null    int64  
 4   genre   350 non-null    object 
dtypes: float64(1), int64(1), object(3)
memory usage: 13.8+ KB

细看分类列

books.value_counts("genre")

genre
Non Fiction    179
Fiction        131
Childrens       40
dtype: int64

.describe() 数值列

books.describe()

         rating        year
count    350.000000  350.000000
mean     4.608571    2013.508571
std      0.226941    3.284711
min      3.300000    2009.000000
25%      4.500000    2010.000000
50%      4.600000    2013.000000
75%      4.800000    2016.000000
max      4.900000    2019.000000

可视化数值数据

import seaborn as sns
import matplotlib.pyplot as plt

sns.histplot(data=books, x="rating")
plt.show()

图书评分的直方图

调整箱宽

sns.histplot(data=books, x="rating", binwidth=.1)
plt.show()

将箱宽设为 0.1 的图书评分直方图

Passons à la pratique !

Python 中的探索性数据分析