pivot python dataframe 行到列

Question

I have a python dataframe like this

       date      col1       col2    val
0   2019-01-01    A         hello   12
1   2019-01-01    A           hi    10
2   2019-01-02    B         hello   5
3   2019-01-02    B           hi    6

I want to convert it to

       date      col1       hello  hi
0   2019-01-01    A           12    10
1   2019-01-02    B            5    6

Answer 1

Use pandas pivot_table() method. You can read more about it here .

The values belong to the val column, the indexes which we will retain are the date and col1 , and we would like to "create" columns out of the col2 col.

df = pd.pivot_table(df, values='val', index=['date', 'col1'],columns=['col2'], aggfunc=np.sum, fill_value=0)

Note that might some values be NaN , the above function will automatically set them to 0 , as per the fill_value parameter.

Input:

import pandas as pd
import numpy as np

a = ['2019-01-01', 'A', 'hello', 12]
b = ['2019-01-01', 'A', 'hi', 10]
c = ['2019-01-02', 'B', 'hello', 5]
d = ['2019-01-02', 'B', 'hi', 6]
df = pd.DataFrame([a, b, c, d], columns=['date', 'col1', 'col2', 'val'])

df = pd.pivot_table(df, values='val', index=['date', 'col1'],columns=['col2'], aggfunc=np.sum, fill_value=0)

print(df)

Output:

col2             hello  hi
date       col1           
2019-01-01 A        12  10
2019-01-02 B         5   6