数据据统计基础之Pandas.DataFrame学习

Pandas 是 Python 中用于数据分析的核心库之一，其中 DataFrame 是最常用的数据结构，类似于 Excel 表格或 SQL 表。下面我们将从 创建、属性、索引、筛选、统计、排序 等多个维度全面介绍 DataFrame 的基本用法。

# dataframe的创建方式importpandasaspdimportnumpyasnp# 通过series来创建s1=pd.Series([1,2,3,4,5])s2=pd.Series([6,7,8,9,10])df=pd.DataFrame({"第一列":s1,"第二列":s2})type(df["第一列"])# 通过字典来创建df=pd.DataFrame({"id":[1,2,3,4,5],"name":["Tom","jack","alice","bob","allen"],"age":[15,17,20,26,30],"score":[60.5,80,30.6,70,83.5]},index=[1,2,3,4,5],columns=["name","id","age","score"])df

# dataframe的属性print('行索引：')print(df.index)print('列标签：')print(df.columns)print('值：')print(df.values)

行索引： Index([1, 2, 3, 4, 5], dtype='int64') 列标签： Index(['name', 'id', 'age', 'score'], dtype='object') 值： [['Tom' 1 15 60.5] ['jack' 2 17 80.0] ['alice' 3 20 30.6] ['bob' 4 26 70.0] ['allen' 5 30 83.5]]

print('维度：',df.ndim)print('数据类型：')print(df.dtypes)print('形状：',df.shape)print('元素个数：',df.size)

维度： 2 数据类型： name object id int64 age int64 score float64 dtype: object 形状： (5, 4) 元素个数： 20

# 行列转置print(df.T)

1 2 3 4 5 name Tom jack alice bob allen id 1 2 3 4 5 age 15 17 20 26 30 score 60.5 80.0 30.6 70.0 83.5

# 获取元素 loc iloc at iat# 某行print(df.loc[4])print(df.iloc[3])

name bob id 4 age 26 score 70.0 Name: 4, dtype: object name bob id 4 age 26 score 70.0 Name: 4, dtype: object

# 某列print(df.loc[:,'name'])print(df.iloc[:,0])

1 Tom 2 jack 3 alice 4 bob 5 allen Name: name, dtype: object 1 Tom 2 jack 3 alice 4 bob 5 allen Name: name, dtype: object

# 单个元素print(df.at[3,'score'])print(df.iat[2,1])print(df.loc[3,'score'])print(df.iloc[2,1])

30.6 3 30.6 3

# 获取单列数据print(df['name'])print(type(df['name']))print(df.name)

1 Tom 2 jack 3 alice 4 bob 5 allen Name: name, dtype: object <class 'pandas.core.series.Series'> 1 Tom 2 jack 3 alice 4 bob 5 allen Name: name, dtype: object

print(df[['name','score']])#多列数据的获取

name score 1 Tom 60.5 2 jack 80.0 3 alice 30.6 4 bob 70.0 5 allen 83.5

# 查看部分数据print(df.head(2))print(df.tail(3))

name id age score 1 Tom 1 15 60.5 2 jack 2 17 80.0 name id age score 3 alice 3 20 30.6 4 bob 4 26 70.0 5 allen 5 30 83.5

# 使用布尔索引筛选数据df[df.score>70]df[(df.score>70)&(df.age<20)]

# 随机抽样df.sample(3)

print(df.isin(['jack',20]))# 查看元素是否包含在参数集合中

name id age score 1 False False False False 2 True False False False 3 False False True False 4 False False False False 5 False False False False

print(df.isna())# 查看元素是否是缺失值

name id age score 1 False False False False 2 False False False False 3 False False False False 4 False False False False 5 False False False False

print(df['score'].sum())# 某一列的总和print(df.score.max())# 最大值print(df.age.min())# 最小值print(df.score.mean())# 平均值print(df.score.median())# 中位数print(df.age.mode())# 众数print(df.score.std())# 标准差print(df.score.quantile(0.25))# 分位数print(df.describe())

324.6 83.5 15 64.92 70.0 0 15 1 17 2 20 3 26 4 30 Name: age, dtype: int64 21.188605428390044 60.5 id age score count 5.000000 5.000000 5.000000 mean 3.000000 21.600000 64.920000 std 1.581139 6.268971 21.188605 min 1.000000 15.000000 30.600000 25% 2.000000 17.000000 60.500000 50% 3.000000 20.000000 70.000000 75% 4.000000 26.000000 80.000000 max 5.000000 30.000000 83.500000

print(df.count())# 返回每一列非缺失值的个数

name 5 id 5 age 5 score 5 dtype: int64

print(df.value_counts())# 出现的次数

name id age score Tom 1 15 60.5 1 alice 3 20 30.6 1 allen 5 30 83.5 1 bob 4 26 70.0 1 jack 2 17 80.0 1 Name: count, dtype: int64

print(df.drop_duplicates())

name id age score 1 Tom 1 15 60.5 2 jack 2 17 80.0 3 alice 3 20 30.6 4 bob 4 26 70.0 5 allen 5 30 83.5

print(df.duplicated(subset=['age']))# 查看是否重复

1 False 2 False 3 False 4 False 5 False dtype: bool

df.sample(2)# 随机抽样

print(df.replace(15,30))

name id age score 1 Tom 1 30 60.5 2 jack 2 17 80.0 3 alice 3 20 30.6 4 bob 4 26 70.0 5 allen 5 30 83.5

df.cumsum()df.cummax()df.cummin(axis=0)

print(df.sort_index(ascending=False))

name id age score 5 allen 5 30 83.5 4 bob 4 26 70.0 3 alice 3 20 30.6 2 jack 2 17 80.0 1 Tom 1 15 60.5

print(df.sort_values(by=['score','age']))

name id age score 3 alice 3 20 30.6 1 Tom 1 15 60.5 4 bob 4 26 70.0 2 jack 2 17 80.0 5 allen 5 30 83.5

df.nlargest(2,columns=['score','age'])df.nsmallest(2,columns=['score','age'])