How to Convert a Pandas DataFrame to a Dictionary with Different Orientations?

Converting a Pandas DataFrame to a Dictionary

To convert a Pandas DataFrame to a dictionary, use the to_dict() method. By default, this method uses the DataFrame's column names as dictionary keys and creates a dictionary of index:data pairs for each column.

df.to_dict()

Customizing the Dictionary Output

To obtain a list of values for each column instead of a dictionary of index:data pairs, use the orient argument. Here are the available orientations:

• dict: Default orientation (column names as keys, index:data pairs as values)
• list: Keys are column names, values are lists of column data
• series: Keys are column names, values are Series objects containing the data
• split: Splits columns/data/index into separate keys
• records: Each row becomes a dictionary with column names as keys and data values as values
• index: Similar to 'records', but keys are index labels instead of a list

Example

Consider the following DataFrame:

df = pd.DataFrame({'ID': ['p', 'q', 'r'], 'A': [1, 4, 4], 'B': [3, 3, 0], 'C': [2, 2, 9]})

To convert this DataFrame to a dictionary with 'ID' as keys and the other columns' values as lists, use the following code:

df.set_index('ID').T.to_dict('list')

This will return the following dictionary:

{'p': [1, 3, 2], 'q': [4, 3, 2], 'r': [4, 0, 9]}

Other Orientations

Here are examples of the different orientations:

dict:

df.to_dict('dict')

Output:

{'ID': {'p': 'p', 'q': 'q', 'r': 'r'},
 'A': {0: 1, 1: 4, 2: 4},
 'B': {0: 3, 1: 3, 2: 0},
 'C': {0: 2, 1: 2, 2: 9}}

list:

df.to_dict('list')

Output:

{'ID': ['p', 'q', 'r'], 'A': [1, 4, 4], 'B': [3, 3, 0], 'C': [2, 2, 9]}

series:

df.to_dict('series')

Output:

{'ID': 0    p
 1    q
 2    r
 Name: ID, dtype: object,
 'A': 0    1
 1    4
 2    4
 Name: A, dtype: int64,
 'B': 0    3
 1    3
 2    0
 Name: B, dtype: int64,
 'C': 0    2
 1    2
 2    9
 Name: C, dtype: int64}

split:

df.to_dict('split')

Output:

{'columns': ['ID', 'A', 'B', 'C'], 'data': [['p', 1, 3, 2], ['q', 4, 3, 2], ['r', 4, 0, 9]], 'index': [0, 1, 2]}

records:

df.to_dict('records')

Output:

[{'ID': 'p', 'A': 1, 'B': 3, 'C': 2}, {'ID': 'q', 'A': 4, 'B': 3, 'C': 2}, {'ID': 'r', 'A': 4, 'B': 0, 'C': 9}]

index:

df.to_dict('index')

Output:

{0: {'ID': 'p', 'A': 1, 'B': 3, 'C': 2},
 1: {'ID': 'q', 'A': 4, 'B': 3, 'C': 2},
 2: {'ID': 'r', 'A': 4, 'B': 0, 'C': 9}}

The above is the detailed content of How to Convert a Pandas DataFrame to a Dictionary with Different Orientations?. For more information, please follow other related articles on the PHP Chinese website!