transform_fold() does not work after renaming columns

Running

d1_coll = alt.hconcat()

d1_bvb = bvb.copy()
d1_bvb.rename(columns = {'value' : 'value_renamed', 'fee' : 'fee_renamed'}, inplace=True)

for chart in [
    [0 , 'All players'], 
    [1, 'Excluding free agents'], 
    [1000000, 'Transfers above €1 million']
    ]: 
    
    d1_chart = alt.Chart(
        d1_bvb,
        title = chart[1]
        
    ).transform_filter(
        datum.fee >= chart[0]
        
    ).transform_fold(
        ['value_renamed', 'fee_renamed'], as_ = ['key_fold', 'value_fold']
        
    ).mark_bar(
        clip=True
    ).encode(
        y=alt.Y('mean(value_fold):Q', scale=alt.Scale(domain=(0,14000000)), title='Average in (€)'),
        x=alt.X('key_fold:N', title = 'x-Axis title'),
        color=alt.Color('key_fold:N'),
        column=alt.Column('trans_type', title='column title', spacing=5)
    )
    
    d1_coll |= d1_chart
    
d1_coll

returns no chart.

When I comment out the d1_bvb.rename() line and change the first argument in transform_fold() to value and fee, it returns the correct chart.

[jakevdp]

The only way to change the names of your columns once they’re passed to the chart is to use a calculate transform. You could do something like this:

import numpy as np
import pandas as pd
import altair as alt

rand = np.random.RandomState(0)
data = pd.DataFrame({
    'date': pd.date_range('2019-01-01', freq='D', periods=30),
    'data_a': rand.randn(30).cumsum(),
    'data_b': rand.randn(30).cumsum(),
    'data_c': rand.randn(30).cumsum(),
})

alt.Chart(data).transform_calculate(**{
    'Data A': 'datum.data_a',
    'Data B': 'datum.data_b',
    'Data C': 'datum.data_c',
}).transform_fold(
    ['Data A', 'Data B', 'Data C'],
).mark_line().encode(
    x='date:T',
    y='value:Q',
    color='key:N',
)

[yannxaver]

Thank you!