to_pandas

The to_pandas method converts a Deephaven table to a pandas.DataFrame.

Caution

to_pandas clones an entire table into memory. For large tables, consider limiting the number of rows and columns in the table before using this method.

Syntax

to_pandas(
  table: Table,
  cols: list[str] = None,
  dtype_backend: str = 'numpy_nullable',
  conv_null: bool = True
) -> pandas.DataFrame

Parameters

ParameterTypeDescription
tableTable

The source table to convert.

cols optionallist[str]

The source table columns to convert. If not specified, all columns are converted.

dtype_backend optionalstr

Which dtype_backend to use. The default is "numpy_nullable".

  • "numpy_nullable" uses nullable NumPy data types as the backend.
  • None uses non-nullable NumPy data types as the backend.
  • "PyArrow" uses PyArrow data types as the backend.

Both “numpy_nullable” and “pyarrow” automatically convert Deephaven nulls to pandas NA and enable pandas extension types. Extension types are needed to support types beyond NumPy’s type system. Extension types support operations such as properly mapping Java Strings to Python strings.

conv_null optionalbool

When dtype_backend is set to None, conv_null determines whether to check for Deephaven nulls in the data and automatically replace them with pd.NA. The default is True.

Returns

A pandas.DataFrame.

Example

The following example uses new_table to create a table, then converts it to a pandas.DataFrame with to_pandas.

from deephaven.pandas import to_pandas
from deephaven import new_table
from deephaven.column import int_col

source = new_table([int_col("Num1", [1, 2, 3, 4]), int_col("Num2", [5, 6, 7, 8])])

result = to_pandas(source)