代码

# check if value is the header/column name
m = df['type'].str.endswith('Product')

# mask and forward fill to associate
# column name with each value below it
df['cols'] = df['type'].mask(~m).ffill()

# pivot to reshape to wide format
result = df[~m].pivot(index='id', columns='cols', values='type')

结果

cols  First Product Second Product
id                                
cust1         chair          table
cust2         chair           desk

您可以尝试使用移位来连接 Dataframe ，并透视结果：

temp = df.join(df.shift(), rsuffix='_2').rename_axis(index='ix').reset_index()
temp = temp.drop(columns=['id_2'])[temp.index % 2 == 1].set_index('ix')
result = temp.pivot('id', 'type_2', 'type').rename_axis(
    index='cust', columns=None).reset_index()

它给出：

cust First Product Second Product
0  cust1         chair          table
1  cust2         chair           desk

使用zip和list comprehension，您可以执行以下操作：

df = pd.DataFrame.from_dict({
  'type': { 4: 'Second Product', 5: 'table', 6: 'First Product', 7: 'chair',
            8: 'Second Product', 9: 'desk', 10: 'First Product', 11: 'chair' },
  'id': { 4: 'cust1', 5: 'cust1', 6: 'cust1', 7: 'cust1',
          8: 'cust2', 9: 'cust2', 10: 'cust2', 11: 'cust2' }
})

itpZip = list(zip(df['id'][::2], df['type'][::2], [*df['type'][1::2], None]))
getProdTypes = lambda c: set(ctp[1] for ctp in itpZip if c==ctp[0])
custProdTypes = {ci: getProdTypes(ci) for ci in set(df['id'][::2])}
df = pd.DataFrame([{
    'cust': ci, **{ti:', '.join(
        p for c,t,p in itpZip if c==ci and t==ti
      ) for ti in getProdTypes(ci)}
} for ci in set(df['id'][::2])])

[ view stages and output ]