pandas 如何使用熊猫计算与起始值相比的百分比变化？
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/35090498/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
How to calculate percent change compared to the beginning value using pandas?
提问by E.K.
I have a DataFrameand need to calculate percent change compared to the beginning of the year by companies. Is there any way to use pct_change()or other method to perform this task? Thanks!
我有一个DataFrame并且需要计算与公司年初相比的百分比变化。有什么方法可以使用pct_change()或其他方法来执行此任务吗？谢谢！
dflooks like
df好像
security date price
IBM 1/1/2016 100
IBM 1/2/2016 102
IBM 1/3/2016 108
AAPL 1/1/2016 1000
AAPL 1/2/2016 980
AAPL 1/3/2016 1050
AAPL 1/4/2016 1070
results I want
我想要的结果
security date price change
IBM 1/1/2016 100 NA
IBM 1/2/2016 102 2%
IBM 1/3/2016 108 8%
AAPL 1/1/2016 1000 NA
AAPL 1/2/2016 980 -2%
AAPL 1/3/2016 1050 5%
AAPL 1/4/2016 1070 7%
回答by Stefan
Sounds like you are looking for an expanding_windowversion of pct_change(). This doesn't exist out of the box AFAIK, but you could roll your own:
像你这样的声音正在寻找一个expanding_window版本pct_change()。这不是开箱即用的 AFAIK，但您可以推出自己的：
df.groupby('security')['price'].apply(lambda x: x.div(x.iloc[0]).subtract(1).mul(100))
回答by fivetentaylor
This works, assuming you're already ordered by date within each possible grouping.
这是有效的，假设您已经在每个可能的分组中按日期排序。
def pct_change(df):
df['pct'] = 100 * (1 - df.iloc[0].price / df.price)
return df
df.groupby('security').apply(pct_change)
回答by Marco
I had the same problem, but solved it his way:
我遇到了同样的问题，但以他的方式解决了：
(only difference was that the columns would be your company and not the row.)
（唯一的区别是列将是您的公司而不是行。）
for each column of my dataframe I did:
对于我的数据框的每一列，我做了：
df[column] = df[column].pct_change().cumsum()
df[column] = df[column].pct_change().cumsum()
pct_change()calculates the change between now and the last value, and cumcum()adds it all together.
pct_change()计算现在和最后一个值之间的变化，cumcum()并将它们加在一起。
