DataFrame to_dict method should also provide orient parameter (like to_json)

[femtotrader]

Hello,

it will be nice if to_dict method could provide same orient parameter as to_json.
For example when outtype='split' we get same results as outtype='series'.

I also noticed that df.to_dict(outtype='split1234') is understood as df.to_dict(outtype='series') which is quite strange but df.to_dict(outtype='a1234') raises ValueError: outtype a1234 not understood which is a correct behavior

Kind regards

Femto

[jreback]

http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.to_dict.html?highlight=to_dict#pandas.DataFrame.to_dict

can you show what kind of mapping you would expect here?

care to submit a pull-request for this?

[femtotrader]

I'm looking for orient='split'.
Not sure that my code can be clean enough to make a PR.
Moreover I wonder if a refactoring between to_json and to_dict is not necessary.

[jreback]

not sure what you are asking? isn't `outtype='series`` what you are asking?

maybe show an example?

I suppose the API is different between to_json and to_dict they have different purposes. What are you proposing?

[femtotrader]

See

In [44]: df
Out[44]:
   c1
0   1
1   2
2   4

In [45]: df.to_json(orient="split")
Out[45]: '{"columns":["c1"],"index":[0,1,2],"data":[[1],[2],[4]]}'

In [46]: df.to_dict(outtype="series")
Out[46]:
{'c1': 0    1
 1    2
 2    4
 Name: c1, dtype: int64}

I'm just proposing to have the same option name: "orient" for both method (but we can have different parameter name... but same is more convenient)
I'm also proposing to provide orient=split

In [46]: json.loads(df.to_json(orient="split"))
Out[47]: {u'columns': [u'c1'], u'data': [[1], [2], [4]], u'index': [0, 1, 2]}

so it will be nice if we could have:

In [46]: json.to_dict(orient="split")
Out[47]: {u'columns': [u'c1'], u'data': [[1], [2], [4]], u'index': [0, 1, 2]}

[jreback]

ic.

ok, would then take a pull-request for this:

• deprecate outtype, rename to orient, pretty straightforward (you deprecate if the outtype is explicity specified, accepting both till the deprecation is removed (in a future version). their are some decorators to do this in other parts of core/frame.py
• add split option

[femtotrader]

my problem is that I don't know how I can have my forked version and also stable version on the same computer.

@deprecate_kwarg(old_arg_name='outtype', new_arg_name='orient')
def to_dict(self, orient='dict'):
    """
    Convert DataFrame to dictionary.

    Parameters
    ----------
    orient : str {'dict', 'list', 'series', 'records', 'split'}
        Determines the type of the values of the dictionary. The
        default `dict` is a nested dictionary {column -> {index -> value}}.
        `list` returns {column -> list(values)}. `series` returns
        {column -> Series(values)}. `records` returns [{columns -> value}].
        `split` returns dict like {index -> [index], columns -> [columns], data -> [values]}.


    Returns
    -------
    result : dict like {column -> {index -> value}}
    """
    if not self.columns.is_unique:
        warnings.warn("DataFrame columns are not unique, some "
                      "columns will be omitted.", UserWarning)
    if outtype.lower().startswith('d') or orient=='dict':
        return dict((k, v.to_dict()) for k, v in compat.iteritems(self))
    elif outtype.lower().startswith('l') or orient=='list':
        return dict((k, v.tolist()) for k, v in compat.iteritems(self))
    elif orient=='split':
        return {'columns': self.columns, 'index': self.index, 'data': self.values.tolist()}
    elif outtype.lower().startswith('s') or orient=='series':
        return dict((k, v) for k, v in compat.iteritems(self))
    elif outtype.lower().startswith('r') or orient=='records':
        return [dict((k, v) for k, v in zip(self.columns, row))
                for row in self.values]
    else:  # pragma: no cover
        raise ValueError("orient %s not understood" % orient)

[jreback]

easy enough, simply clone the repo complete instruction are here: https://github.com/pydata/pandas/wiki

tests first, then a code change.