Create column from another column in which each row indicates the column to use to obtain that cell

Question

Create column from another column in which each row indicates the column to use to obtain that cell

Navigation

#1 by (1 votes)

3

I have the following information in a DataFrame in Python:

##    x1   x2   x3   x4   x5   colum
##0  206  214  021  122  554     2
##1  226  234  123  456  789     4
##2  245  253  558  855  123     5
##3  265  272  000  111  222     4
##4  283  291  214  589  996     1

and I need to generate a new column depending on the value that contains the colum column, as follows:

##    x1   x2   x3   x4   x5   colum   newColum
##0  206  214  021  122  554     2       214
##1  226  234  123  456  789     4       456
##2  245  253  558  855  123     5       123
##3  265  272  000  111  222     4       111
##4  283  291  214  589  996     1       283

I do not know if I'm clear on my application, I thank whoever can help me.

python python-3.x pandas

asked by Yeison H. Arias 12.09.2018 в 12:36

source

1 answer

Failed to connect WebSocket php - javascript Get procedure OUTPUT and save in a txt

score 1 · Accepted Answer

You can use pandas.DataFrame.apply on the rows (axis 1) obtaining the name of the column by simply formatting strings:

import io
import pandas as pd

data = io.StringIO("""\
 ##   x1   x2   x3   x4   x5   colum
##0  206  214  021  122  554       2
##1  226  234  123  456  789       4
##2  245  253  558  855  123       5
##3  265  272  000  111  222       4
##4  283  291  214  589  996       1
""")

df = pd.read_table(data, sep="\s+", engine="python",  index_col=0)
df["newColum"] = df.apply(lambda row: row[f'x{row["colum"]}'],  axis=1)

>>> df

      x1   x2   x3   x4   x5  colum  newColum
##                                           
##0  206  214   21  122  554      2       214
##1  226  234  123  456  789      4       456
##2  245  253  558  855  123      5       123
##3  265  272    0  111  222      4       111
##4  283  291  214  589  996      1       283

If there is a possibility that there are values in the column that do not correspond to the head of any column you can use pandas.Series.get and assign NaN for example in these cases:

lambda row: row.get(f'x{row["colum"]}', default=np.nan)

If you use a Python version less than 3.6 you can use str.format ( "x{}".format(row["colum"]) ) instead of string literals formatted ( f'x{row["colum"] ):