Understanding the Transform Function in Pandas

Understanding the Transform Function in Pandas

来源

What is transform?

我在 Python Data Science Handbook 的书中找到了关于这个话题的说明。正如书中所描述的，transform 是一个和groupby同时使用的操作。我推测大多数的pandas用户可能已经用过了aggregate, filter 或者 apply在使用 groupby的同时。然而，transform 有点难以理解。

aggregation会返回数据的缩减版本，而transformation能返回完整数据的某一变换版本供我们重组。这样的transformation，输出的形状和输入一致。一个常见的例子是通过减去分组平均值来居中数据。

First Approach - Merging

data_str='''account,name,order,sku,quantity,unit price,ext price
383080,Will LLC,10001,B1-20000,7,33.69,235.83
383080,Will LLC,10001,S1-27722,11,21.12,232.32
383080,Will LLC,10001,B1-86481,3,35.99,107.97
412290,Jerde-Hilpert,10005,S1-06532,48,55.82,2679.36
412290,Jerde-Hilpert,10005,S1-82801,21,13.62,286.02
412290,Jerde-Hilpert,10005,S1-06532,9,92.55,832.95
412290,Jerde-Hilpert,10005,S1-47412,44,78.91,3472.04
412290,Jerde-Hilpert,10005,S1-27722,36,25.42,915.12
218895,Kulas Inc,10006,S1-27722,32,95.66,3061.12
218895,Kulas Inc,10006,B1-33087,23,22.55,518.65
218895,Kulas Inc,10006,B1-33364,3,72.3,216.9
218895,Kulas Inc,10006,B1-20000,-1,72.18,-72.18'''

import io
import pandas as pd
data=pd.read_csv(io.StringIO(data_str))
order_total = data.groupby('order')['ext price'].sum().rename('order total').reset_index()
data_merge=data.merge(order_total)
data_merge['Percnet Order']=data_merge['ext price']/data_merge['order total']

what is happening with the standard groupby

Second Approach - Using Transform

order_total=data.groupby('order')['ext price'].transform('sum')
data['percent order'] = data['ext price']/order_total

what is happening in transform