CODEWITHSUNDEEP
pythonpandas
polka
polka

Reputation: 1529

combine data in pandas

I have a pandas dataframe like this:

index   integer_2_x  integer_2_y
0        49348          NaN
1        26005          NaN
2            5          NaN
3          NaN           26
4        26129          NaN
5          129          NaN
6          NaN           26
7          NaN           17
8        60657          NaN
9        17031          NaN

I want to make a third column that looks like this by taking the numeric value in the first and the second and eliminating the NaN. How do I do this?

index   integer_2_z
0        49348
1        26005
2            5
3           26
4        26129
5          129
6           26
7           17
8        60657
9        17031

Upvotes: 0

Views: 133

Answers (3)

Romain
Romain

Reputation: 21888

Maybe you can simply use the fillna function.

# Creating the DataFrame
df = pd.DataFrame({'integer_2_x': [49348, 26005, 5, np.nan, 26129, 129, np.nan, np.nan, 60657, 17031],
               'integer_2_y': [np.nan, np.nan, np.nan, 26, np.nan, np.nan, 26, 17, np.nan, np.nan]})

# Using fillna to fill a new column
df['integer_2_z'] = df['integer_2_x'].fillna(df['integer_2_y'])

# Printing the result below, you can also drop x and y columns if they are no more required
print(df)

   integer_2_x  integer_2_y  integer_2_z
0        49348          NaN        49348
1        26005          NaN        26005
2            5          NaN            5
3          NaN           26           26
4        26129          NaN        26129
5          129          NaN          129
6          NaN           26           26
7          NaN           17           17
8        60657          NaN        60657
9        17031          NaN        17031

Upvotes: 0

Ale
Ale

Reputation: 1345

I used http://pandas.pydata.org/pandas-docs/stable/basics.html#general-dataframe-combine

import pandas as pd
import numpy as np
df = pd.read_csv("data", sep="\s*")  # cut and pasted your data into 'data' file
df["integer_2_z"] = df["integer_2_x"].combine(df["integer_2_y"], lambda x, y: np.where(pd.isnull(x), y, x))

Output

       index  integer_2_x  integer_2_y  integer_2_z
0      0        49348          NaN        49348
1      1        26005          NaN        26005
2      2            5          NaN            5
3      3          NaN           26           26
4      4        26129          NaN        26129
5      5          129          NaN          129
6      6          NaN           26           26
7      7          NaN           17           17
8      8        60657          NaN        60657
9      9        17031          NaN        17031

Upvotes: 0

Jianxun Li
Jianxun Li

Reputation: 24742

One way is to use the update function.

import pandas as np
import numpy as np

# some artificial data
# ========================
df = pd.DataFrame({'X':[10,20,np.nan,40,np.nan], 'Y':[np.nan,np.nan,30,np.nan,50]})
print(df)


    X   Y
0  10 NaN
1  20 NaN
2 NaN  30
3  40 NaN
4 NaN  50    

# processing
# =======================
df['Z'] = df['X']
# for every missing value in column Z, replace it with value in column Y
df['Z'].update(df['Y'])
print(df)

    X   Y   Z
0  10 NaN  10
1  20 NaN  20
2 NaN  30  30
3  40 NaN  40
4 NaN  50  50

Upvotes: 1

Related Questions