1. print(pandas.__version__)
终端conda update pandas
2.
df = pd.read_csv("NBA Player.csv",sep = "\t")
3.
——
a = {"name":"xiaoming","age":20,"district":Beijin}
pd.Series(a)
——
b = [1,2,3,4,5]
pd.Series(b,index =list( "abcde"))
4.
——
a = [[1,2,3],[7,8,9]]
pd.Dataframe(a,columns = list("ab"),index = list("ABC"))
5.
df.columns #列名
df.index #索引
df.dtype #返回每一列的数值类型（非数值都返回object）
df.shape
df.zise
len(df)
df.head()
df.tail()
df.rename = (columns = {"height":"Height"},inplace = True) #inplace = True 修改原数据集
df.replace = ("Height":{A:B})
df.collage.value_counts() #计数
df.sort_values(by = ["Height","Weight"]) 根据某些列值排序
df.describe #描述的是数据集里面所有的特征数
df.max(axis= 0) #默认为axis = 0,针对列计算
6. #数据选取、添加、删除
——
df[["Height","Weight"]]
df.Height
——
df.["glass"] = 1 #每次只能增加一列
——
df[(df["Height"] > 200) | (df["Height"] < 50)]
——
del df["glass"] #对本身做修改
7. #缺失值
pd.isnull(df)
pd.isnull(df["Height"])
df.dropna(axis =0,how = "any", thresh = None, inplace = False)
df.fillna(value = None,method = None,axis = None,inplace = False, limit = None, downcast = None,**kwargs)
8. #文本数据
——
s = pd.Series(["AB","abc","efg"])
s.str.strip().str.endwith("o")
——
df["Player"].str.split(" ").str.get(1)
df["Player"].split(" ",expand = True)
df["Player"].str[:3]
9.#索引操作
——
df[:5] #返回5行
——
df.loc[:5] #返回6行,loc基于lable，
df.loc[["Arai","Kalif"]]
df.loc[df[["Arai","Kalif"],["Height"]>= 180]]
——
df.iloc[:5] #返回5行,基于位置
——
df.loc[(df['Height']>= 180) &(df['weight']>= 80),'flag'] = 'high'
df.loc[(df['Height']>= 180) &(df['weight']>= 170),'flag'] = 'msize'
df.loc[~(df['Height']>= 180) &(df['weight']>= 170),'flag'] = 'small'
10. #分组计算
grouped = df.groupby('director_name')
grouped.mean()
grouped['duration'].mean()
grouped.std()
import numpy as np
grouped.agg([np.mean,np.std])
grouped.agg({'duration':np.mean,'facebook_like':np.mean'})
11.#transformation 标准化
z_score = lambda s: (s-s.mean())/s.std()
grouped[['facebook_like','duration']].transformation(z_sccore)
12.#Filteration过滤
grouped.filter(lambda g: len(g)>1)['daration'].value_counts()
13.#表连接
result = concat([df1,df2,df3])
pd.merge(left,right,on = 'key')
