Python Forum
pandas dataframe iloc mystery - Printable Version

+- Python Forum (https://python-forum.io)
+-- Forum: Python Coding (https://python-forum.io/forum-7.html)
+--- Forum: Data Science (https://python-forum.io/forum-44.html)
+--- Thread: pandas dataframe iloc mystery (/thread-21912.html)

pandas dataframe iloc mystery - edvvardbrian - Oct-20-2019

Hello!

So I have a pandas dataframe filled with zeros, and I want to set the very first column and the very first row to the following values: ["a_1","a_2","a_3","b_4","b_5","b_6"].

I'm using this code:

import numpy as np
import pandas as pd
data_array = np.zeros([7, 7])

data_df = pd.DataFrame(data_array, index=["indexes","a","a","a","b","b","b"],
                                columns=["columns","a","a","a","b","b","b"])

print(data_df, "\n")

data_df.iloc[:,0] = [0] + ["a_1","a_2","a_3","b_4","b_5","b_6"]

print(data_df, "\n")

data_df.iloc[0,:] = [0] + ["a_1","a_2","a_3","b_4","b_5","b_6"]

print(data_df, "\n")
However, this does not work, because the final output is:

Output:
 columns    a    a    a    b    b    b
indexes      0.0  0.0  0.0  0.0  0.0  0.0  0.0
a            0.0  0.0  0.0  0.0  0.0  0.0  0.0
a            0.0  0.0  0.0  0.0  0.0  0.0  0.0
a            0.0  0.0  0.0  0.0  0.0  0.0  0.0
b            0.0  0.0  0.0  0.0  0.0  0.0  0.0
b            0.0  0.0  0.0  0.0  0.0  0.0  0.0
b            0.0  0.0  0.0  0.0  0.0  0.0  0.0 

        columns    a    a    a    b    b    b
indexes       0  0.0  0.0  0.0  0.0  0.0  0.0
a           a_1  0.0  0.0  0.0  0.0  0.0  0.0
a           a_2  0.0  0.0  0.0  0.0  0.0  0.0
a           a_3  0.0  0.0  0.0  0.0  0.0  0.0
b           b_4  0.0  0.0  0.0  0.0  0.0  0.0
b           b_5  0.0  0.0  0.0  0.0  0.0  0.0
b           b_6  0.0  0.0  0.0  0.0  0.0  0.0 

        columns    a    a    a    b    b    b
indexes       0  a_3  a_3  a_3  b_6  b_6  b_6
a           a_1    0    0    0    0    0    0
a           a_2    0    0    0    0    0    0
a           a_3    0    0    0    0    0    0
b           b_4    0    0    0    0    0    0
b           b_5    0    0    0    0    0    0
b           b_6    0    0    0    0    0    0
As you can see in the last table, the first column is set to the right values (a_1 a_2 a_3 b_4 b_5 b_6), but the first row is set instead to (a_3 a_3 a_3 b_6 b_6 b_6).

For the life of me I can't figure out why this happens.

Thanks!

RE: pandas dataframe iloc mystery - jefsummers - Oct-28-2019

What if you change the names of the columns to 6 different letters? Thinking about what goes on under the hood with keys and lookups I can see Pandas getting confused by identically named columns.

RE: pandas dataframe iloc mystery - jefsummers - Oct-29-2019

I just tested and that fixed it. Moral of the story - do not duplicate names of columns.