Python，Pandas数据框语法

Question

I have a data frame(df) and I am trying to copy first line from df to df1 by the following syntax.

df1=df.iloc[0].copy()

and when I printed df1 the data frame look like this

device_consumption_key                2700848
eagle_id                         d8d5b9008b92
device_id                    002446000006983b
component_id                               0b
read_dt                   2017-10-25 20:05:24
msg_dt                    2017-09-15 14:09:19
delivered                                0.86
received                                    0
unit                                      kWh
creation_time             2017-10-25 20:06:00
etl_pid                        20171102184518
Name: 0, dtype: object

if I used a different syntax liek below the frame looks like a da-->

df1=df.iloc[[0]].copy()

Can't able to show the full picture but I guess you got the idea

Why is that?

Answer 1

The two difference slicing methods, using double brackets and single brackets return a DataFrame and a Series respectively:

>>> type(df.iloc[[0]].copy())
<class 'pandas.core.frame.DataFrame'>
>>> type(df.iloc[0].copy())
<class 'pandas.core.series.Series'>

Series are by default displayed with the column names as the index, and the values as what looks like a column , but should not be considered as such.

More details :

A Series , based on the pandas documentation is a

One-dimensional ndarray with axis labels (including time series).

Whereas a DataFrame

Can be thought of as a dict-like container for Series objects

Answer 2

Take a look at the .loc documentation , specifically (highlights added by me)

Single label. Note this returns the row as a Series.

df.loc['viper']

And

List of labels. Note using [[]] returns a DataFrame.

df.loc[['viper', 'sidewinder']]

Basically when you slice using a single label, such as df.iloc[0] , you get a pd.Series object. When you slice using a list of labels , such as [[0]] , you get a pd.DataFrame .

The same logic applies to .iloc

---

As highlighted in comments, this is assuming unique indexes . If you use loc in a data frame that has duplicated indexes, then unexpected results might arise (which is one strong argument for you not to use duplicated indexes !)