How to turn a pandas pivot_table into a simple table - pandas

Question

I have a simple DataFrame defined by:
df = pd.DataFrame({
'year': [2019, 2020, 2019, 2020, 2019, 2020, 2019, 2020],
'name': ['Alice', 'Alice', 'Alice', 'Alice', 'Bob', 'Bob', 'Bob', 'Bob'],
'sales': [100, 200, 300, 400, 500, 600, 700, 800]
})
This DataFrame is easily turned into a pivot table using pivot_table :
table = pd.pivot_table(
df,
index=['name'],
columns=['year'],
aggfunc=np.sum)
Now I need to turn this DataFrame into a simple JSON array. Unfortunately, the to_json method doesn't return a simple array:
table.reset_index().to_json(orient="records")
[
{
"["name",""]":"Alice",
"["sales",2019]":400,
"["sales",2020]":600
},
{
"["name",""]":"Bob",
"["sales",2019]":1200,
"["sales",2020]":1400}
]
How can I turn the table DataFrame into a simple (without multiindex) DataFrame?
[
{
"name":"Alice",
"2019":400,
"2020":600
},
{
"name":"Bob",
"2019":1200,
"2020":1400
}
]

Answer 1

you need to pass the values param to get rid of the multiindex , also get rid of the lists in the param inside pivot when passing a single column:
table = pd.pivot_table(
df,
index='name',
columns='year',
values='sales',
aggfunc=np.sum)
table.reset_index().to_json(orient="records")
'[{"name":"Alice","2019":400,"2020":600},{"name":"Bob","2019":1200,"2020":1400}]'
Adding another alternative if you like:
out = (df.groupby(['name','year'])['sales'].sum().unstack()
.reset_index().to_json(orient='records'))
'[{"name":"Alice","2019":400,"2020":600},{"name":"Bob","2019":1200,"2020":1400}]'