CODEWITHSUNDEEP
pythonpandas
TacoCat
TacoCat

Reputation: 469

Pandas .loc[] method only returns only DataType not Series despite calling single index

TL;DR: .loc[] returns DataFrame type all the time. Even when specifying single index.

I've tried everything. This is driving me insane.

I can't seem to reproduce it anywhere else.

I've checked every Type of data that's beeing passed. Everything is as it should be. But no matter what I pass into the .loc[] it will return a DataFrame not series. 

import numpy as np
import pandas as pd
import datetime


index_list = 'A B C D E F G H'.split()
df = pd.DataFrame(data=None,index=index_list)

k = 0
while k <= 2:
    now = datetime.datetime.now().strftime('%H:%M:%S')
    df.loc[:,now] = 1

    for i in index_list:
        print(df.loc[i])
        print(type(df.loc[i]))
    k += 1

The code above will run with 0 errors and return Series Type of data. This is distilled code but it's exactly the same as the real one. Same flow, exaclty the same Type of data is beeing passed.

  1. 'now' is set as a column name and all the new values for every index is now 1.
  2. Next the script iterates trough the index_list and prints the type.

The problem is that in the real script .loc will only return DataFrame type not Series. And I have no idea why. I even tried to manually enter the .loc index name to check if I'm not passing the wrong type of data. Still returned DataFrame.

I'm 100% out of ideas of what I could be doing wrong.

Maybe some of you have ideas?

EDIT

removed The original code.

I found that if I call print(df.loc[i].iloc[0]) It will return the Series data for the column. 

print(type(df.loc[i].iloc[0]))

Will print:

20:48:48    1
Name: (A,), dtype: int64
<class 'pandas.core.series.Series'>

Why is the name (A,) a tuple?

Upvotes: 1

Views: 1312

Answers (1)

DSM
DSM

Reputation: 353149

TLDR: remove your extra brackets when building dfCoinMaster's index.

In the working code:

df = pd.DataFrame(data=None,index=index_list)

In the non-working code:

dfCoinMaster = pd.DataFrame(data=None,index=[current_coin_listings])

You're adding an extra level of list nesting, which you can see in your

Name: (A,), dtype: int64

line. You can reproduce the same behaviour by adding the extra brackets to your test:

In [28]: df = pd.DataFrame(data=None, index=[index_list])

In [29]: df.loc[:, 'test'] = 10

In [30]: df
Out[30]: 
   test
A    10
B    10
C    10
D    10
E    10
F    10
G    10
H    10

In [31]: df.loc['A']
Out[31]: 
   test
A    10

In [32]: type(_)
Out[32]: pandas.core.frame.DataFrame

But:

In [33]: df.loc[('A',)]
Out[33]: 
test    10
Name: (A,), dtype: int64

In [34]: type(_)
Out[34]: pandas.core.series.Series

Upvotes: 1

Related Questions