我已使用pandas_datareader将来自Yahoo的股票数据导入数据框。 有两列:日期和股票的调整后收盘价。
Date Adj Close
2017-08-31 168.851196
2017-09-01 169.867691
2017-09-05 165.333496
2017-09-06 165.233810
2017-09-07 166.001160
2017-09-08 163.121201
2017-09-11 168.412735
2017-09-12 169.020630
2017-09-13 169.777969
2017-09-14 168.811356
2017-09-15 179.484131
2017-09-18 186.898300
2017-09-19 186.698990
2017-09-20 185.194214
2017-09-21 180.131882
2017-09-22 178.377991
2017-09-25 170.405807
2017-09-26 171.362473
2017-09-27 175.119354
2017-09-28 175.069534
2017-09-29 178.148788
2017-10-02 178.377991
2017-10-03 178.746704
2017-10-04 180.241486
2017-10-05 180.141861
2017-10-06 180.670013
2017-10-09 184.745804
2017-10-10 188.273499
2017-10-11 190.276505
2017-10-12 190.366211
我希望能够插入另一列名为“ Log return”的列,该列采用当前日期的Adj关闭时间(由于交易日,日期不全为1天),并将其除以前几天的Adj Close然后取该商的自然对数
即Ln(A(今天)/ A(昨天)),其中A只是adj关闭。
顺便说一下,我的dataframe变量叫df。
import pandas as pd
import pandas_datareader as web
#import datetime internal datetime module
#datetime is a Python module
import datetime
#datetime.datetime is a data type within the datetime module
start = datetime.datetime(2015, 9, 1)
end = datetime.datetime(2018, 12, 31)
#DataReader method name is case sensitive
df = web.DataReader("nvda", 'yahoo', start, end)
#invoke to_csv for df dataframe object from
#DataReader method in the pandas_datareader library
#..\first_yahoo_prices_to_csv_demo.csv must not
#be open in another app, such as Excel
df = df.iloc[0:, 5:] #Trims the set to Adj Close
到目前为止,这就是我的代码。 编辑我不希望A(今天)/ A(昨天)-1,实际上我需要Ln(A(今天)/ A(昨天))。 (自然对数)
答案 0 :(得分:3)
尝试一下:
df['Adj Yesterday'] = df['Adj Close'].shift()
df['Log Return'] = df['Adj Close'] / df['Adj Yesterday'] - 1.
如果这不是您想要的,而是关闭,here is the docs for shift。
如果缺少时态数据,也可以将resample或set_index与date_range一起使用。
答案 1 :(得分:2)
df['Log Return'] = df['Adj Close'].pct_change()
如果需要ln
:
df['Log Return'] = np.log(df['Adj Close'].pct_change())
答案 2 :(得分:1)
您可以尝试以下方法:
# First ensure dates are in order
df = df.sort_values('Date')
# Divide all rows by their previous and find log
diff = np.log(df[1:]['Adj Close'] / df[0:-1]['Adj Close'])
# Add new column, first row will be NaN as it has no previous day
df['Log Return'] = pd.concat(pd.Series([pd.nan]), diff)