Use pandas pivot_table() to convert attribute-value pairs to table

问题

I have a set of attribute,value pairs like this:

date,01-01-2018
product,eggs
price, 5
date,01-10-2018
product,milk
price,3

And I want to create a table like

date,product,price
01-01-2018,eggs,5
01-10-2018,milk,3

I've tried adding headers 'attributes' and attribute_values', creating an arbitrary column "values" and using

pd.pivot_table(av_pairs, index="value", columns=av_pairs.attributes, values=av_pairs.attribute_values)

The error is

pandas.core.base.DataError: No numeric types to aggregate

Using stack() also didn't work. Can anyone point out what I'm looking for?

回答1:

Use pandas.pivot and count new indices by cumcount:

print (df)
         a           b
0     date  01-01-2018
1  product        eggs
2    price           5
3     date  01-10-2018
4  product        milk
5    price           3

df1 = pd.pivot(index=df.groupby('a').cumcount(),
              columns=df['a'],
              values=df['b'])

Or use set_index with unstack:

df1 = df.set_index([df.groupby('a').cumcount(), 'a'])['b'].unstack()

print (df1)
a        date price product
0  01-01-2018     5    eggs
1  01-10-2018     3    milk

来源：https://stackoverflow.com/questions/53190981/use-pandas-pivot-table-to-convert-attribute-value-pairs-to-table