使用Numpy数组更新Pandas数据框中的列的一部分

Question

I have data in a format close to that of df (shown below). My problem now is that I want to populate the data in avg_value with the average value for the past "days_back" days.

import numpy as np
import pandas as pd

df = pd.DataFrame({ 'DAY': np.append(np.ones(24),
                                 [np.multiply(np.ones(24), 2),
                                  np.multiply(np.ones(24), 3),
                                  np.multiply(np.ones(24), 4)]),
                'value': np.random.randn(1, 24*4)[0],
                'avg_value': 0.},
                index=pd.date_range('20150101', periods=24*4, freq="H"))

print(df.tail())
                     DAY  avg_value     value
2015-01-04 19:00:00  4.0        0.0  0.685153
2015-01-04 20:00:00  4.0        0.0  0.670713
2015-01-04 21:00:00  4.0        0.0 -0.519541
2015-01-04 22:00:00  4.0        0.0  0.795619
2015-01-04 23:00:00  4.0        0.0 -0.150966

Coming from R, this would be an easy thing to do.. But when I try to do

df.loc[df["DAY"] == the_day_I_want].avg_value = my_numpy_array

I get

A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

So, as the good boy I am, I proceed with the following

index_row = df.loc[df["DAY"] == the_day_I_want].index
index_col = df.columns.get_loc("avg_value")
df.loc[index_row, index_col] = my_numpy_array

But I still end up with the same error! I bet there is a real easy solution to this problem but I just can't find it :/ Any help would be much appreciated!

Answer 1

You are really close, need specify column in loc :

df.loc[df["DAY"] == the_day_I_want].avg_value = my_numpy_array

what is same as:

df.loc[df["DAY"] == the_day_I_want]['avg_value'] = my_numpy_array

change to:

df.loc[df["DAY"] == the_day_I_want, 'avg_value'] = my_numpy_array

And why need it better explain returning a view versus a copy