<!doctype html>
numpy, pandas do data cleansing
Cleaning numpy, pandas null
There are two ways
- Delete rows of data where the null value
- The missing rows or columns to delete
Using data determination method:
isnull: to determine whether the data is empty, returns True if empty, otherwise it returns False
notnull: to determine whether the data is not empty, if empty returns False, otherwise True
any: Analyzing combination with isnull
dropna: delete data in the empty rows or columns of data, axis and other parameters indicate the contrary, the behavior 0, as 1
fillna: filling parameter data for operation method = 'ffill' representing fill forward, 'bfill' represents rearwardly filled
Examples of an embodiment:
method one:
from pandas import DataFrame,Series
df = DataFrame(data = np.random.randint(0,100,size=(7,5)))
#创建一个7行5列的二维数组
#随机取值从0到100，形式是7行5列的数组
#设置是三个空值
df.iloc[3,4] = None #三行四列的值为空
df.iloc[2,2] = np.nan #设置2行2列的值为NAN
df.iloc[5,3] = None #设置5行3列的值为空
df #panads会自动将None的空值转换成NaN
#清洗空值的两种方式
#方式一删除空所在的行数据
#isnull、notnull、any、all
df.isnull() #用于判断数组内的数据是否为空，如果为空放回True,否则返回False
df.isnull().all(axis=1) #1表示行，0表示列 只有在drop中于此相反
#all是行或列中如果出现False就返回False,只有都是True才返回True
#any是行或者列中如果有一个为True，就返回True
df.isnull().any(axis=1) #1是行，0是列
#将布尔值作为原数据的行索引:保留为True的行数据
df.loc[df.isnull().any(axis=1)] #根据isnull()的判断将有空值的行数据保留
drop_index = df.loc[df.isnull().any(axis=1)].index #提取出存在空值的行索引
df.drop(labels=drop_index,axis=0) #删除所在的行
Method Two:
df.notnull().all(axis=1) #notnull是判断不为空的返回True,否则返回False
#找出所有有空的行数据
#将布尔值作为行索引
df.loc[df.notnull().all(axis=1)]
#根据notnull的判断进行过滤出不为空的行数据
Examples of two ways:
#方式二：dropna:可以直接将缺失的行或者列进行删除
df.dropna(axis=0) #在dropna中0表示行，1表示列
drop_duplications(keep=False)
Remove duplicate rows of data
keep=first
Reservations first row of data, delete other rows
keep=last
Retain the last row of data, delete the other duplicate data