Pandas DataFrame | to_dict method
Start your free 7-days trial now!
Pandas DataFrame.to_dict(~) method converts the DataFrame to a Python dictionary or list of dictionaries.
Parameters
1. orient | string | optional
The rule by which to perform the conversion.
When orient="dict" (default):
{ column_1: index_1: value_1 index_2: value_2 column_2: ...}
When orient="list":
{ column_1: [value_1, value_2, ...] column_2: [...] ...}
When orient="series":
{ column_1: Series(value_1, value_2, ...) column_2: Series(...) ...}
When orient="split":
{ "index": [index_1, index_2, ...] "columns": [column_1, column_2, ...] "data": [value_1, value_2, ...]}
When orient="records":
[ { column_1: value_1 column_2: value_2 ... },{ column_1: value_6 ... }, ...]
When orient="index":
{ index_1: { column_1: value column_2: value ... }, index_2: { column_6 ... }, ...}
2. into | class | optional
The collections.abc.Mapping subclass or its instance. This is useful when, instead of the standard Python dict, you want to convert to another type of dictionary such as collections.defaultdict.
Return Value
Depending on the specified orient and into, the return type will vary.
Examples
Consider the following DataFrame:
df = pd.DataFrame({"A":[2,3],"B":[4,5]}, index=["a","b"])df
A Ba 2 4b 3 5
Specifying orient
dict
df.to_dict()
{'A': {'a': 2, 'b': 3}, 'B': {'a': 4, 'b': 5}}
list
df.to_dict(orient="list")
{'A': [2, 3], 'B': [4, 5]}
series
df.to_dict(orient="series")
{'A': a 2 b 3 Name: A, dtype: int64, 'B': a 4 b 5 Name: B, dtype: int64}
For your reference, we show the df here again:
df
A Ba 2 4b 3 5
split
df.to_dict(orient="split")
{'index': ['a', 'b'], 'columns': ['A', 'B'], 'data': [[2, 4], [3, 5]]}
records
df.to_dict(orient="records")
[{'A': 2, 'B': 4}, {'A': 3, 'B': 5}]
index
df.to_dict(orient="index")
{'a': {'A': 2, 'B': 4}, 'b': {'A': 3, 'B': 5}}
Initialising an ordered dictionary
To initialise an ordered dictionary using its class definition:
from collections import OrderedDictdf.to_dict(into=OrderedDict)
OrderedDict([('A', OrderedDict([('a', 2), ('b', 3)])), ('B', OrderedDict([('a', 4), ('b', 5)]))])
Initialising an defaultdict
Unlike ordered dictionaries, the defaultdict must be initialised before we pass it into to_dict(~):
from collections import defaultdictmy_default_dict = defaultdict(list)df.to_dict(into=my_default_dict)
defaultdict(list, {'A': defaultdict(list, {'a': 2, 'b': 3}), 'B': defaultdict(list, {'a': 4, 'b': 5})})
