NVIDIA cuDF

https://docs.nvidia.com/cudf/latest/
Admin
NVIDIA cuDF is a GPU-accelerated library for tabular data processing that provides a pandas-like...
Tokens:534,497
Snippets:4,921
Trust Score:10
Update:3 days ago
Show doc for...
Tokens:
Raw
### DataFrame setup

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.astype/index.html

Initial setup of a DataFrame for casting examples.

```python
>>> import cudf
>>> df = cudf.DataFrame({'a': [10, 20, 30], 'b': [1, 2, 3]})
>>> df
    a  b
0  10  1
1  20  2
2  30  3
>>> df.dtypes
a    int64
b    int64
dtype: object
```

--------------------------------

### DataFrame setup for loc examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.loc/index.html

Initial state of the DataFrame used in subsequent examples.

```python
>>> df
   a  b
a  0  5
b  1  6
c  2  7
d  3  8
e  4  9
```

--------------------------------

### Setup rolling window examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.rolling/index.html

Initializes the cuDF Series used for subsequent rolling window demonstrations.

```python
>>> import cudf
>>> import numpy as np
>>> import pandas as pd
>>> a = cudf.Series([1, 2, 3, None, 4])
```

--------------------------------

### Setup Series for padding examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.pad

Initializes a cudf Series containing strings to be used in subsequent padding examples.

```python
>>> import cudf
>>> s = cudf.Series(["caribou", "tiger"])
```

--------------------------------

### StringMethods.index with start and end parameters

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.index

This example shows how to use the start and end parameters to restrict the search range for the substring within each string.

```python
>>> s = cudf.Series(['abc', 'abb','ab' ,'ddb'])
>>> s.str.index('b', start=1, end=5)
0    1
1    1
2    1
3    2
dtype: int64
```

--------------------------------

### Create a DataFrame for info examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.info/index.html

Initializes a sample DataFrame with integer, string, and float columns to be used in subsequent examples.

```python
>>> import cudf
>>> int_values = [1, 2, 3, 4, 5]
>>> text_values = ['alpha', 'beta', 'gamma', 'delta', 'epsilon']
>>> float_values = [0.0, 0.25, 0.5, 0.75, 1.0]
>>> df = cudf.DataFrame({"int_col": int_values,
...                     "text_col": text_values,
...                     "float_col": float_values})
>>> df
   int_col text_col  float_col
0        1    alpha       0.00
1        2     beta       0.25
2        3    gamma       0.50
3        4    delta       0.75
4        5  epsilon       1.00
```

--------------------------------

### Initialize DataFrame for string replacement

Source: https://docs.nvidia.com/cudf/latest/cudf/missing-data

Setup a dictionary and convert it into a cuDF DataFrame for string replacement examples.

```python
d = {"a": list(range(4)), "b": list("ab.."), "c": ["a", "b", None, "d"]}
```

```python
df = cudf.DataFrame(d)
```

```python
df
```

--------------------------------

### DataFrame setup

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.drop/index.html

Initialize a DataFrame for demonstration purposes.

```python
>>> import cudf
>>> df = cudf.DataFrame({"A": [1, 2, 3, 4],
...                      "B": [5, 6, 7, 8],
...                      "C": [10, 11, 12, 13],
...                      "D": [20, 30, 40, 50]})
>>> df
   A  B   C   D
0  1  5  10  20
1  2  6  11  30
2  3  7  12  40
3  4  8  13  50
```

--------------------------------

### Install tpchgen-cli

Source: https://docs.nvidia.com/cudf/latest/cudf_pandas/benchmarks

Install the Rust-based TPC-H data generator.

```bash
pip install tpchgen-cli
```

--------------------------------

### Setup PyArrow Decimal Type

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.dtypes.Decimal128Dtype.from_arrow

Initializes the necessary imports and creates a pyarrow decimal128 type with specified precision and scale for use in subsequent examples.

```python
>>> import cudf
>>> import pyarrow as pa
>>> pa_type = pa.decimal128(precision=9, scale=2)
```

--------------------------------

### Check for year start dates

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.series.DatetimeProperties.is_year_start

This example demonstrates how to use the is_year_start property on a datetime Series to identify dates that fall on January 1st.

```python
>>> import pandas as pd, cudf
>>> s = cudf.Series(pd.date_range("2017-12-30", periods=3))
>>> dates
0   2017-12-30
1   2017-12-31
2   2018-01-01
dtype: datetime64[ns]
>>> dates.dt.is_year_start
0    False
1    False
2    True
dtype: bool
```

--------------------------------

### Check for quarter start dates

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.series.DatetimeProperties.is_quarter_start

This example demonstrates how to use the is_quarter_start property on a cuDF Series containing datetime objects to identify the first day of a quarter.

```python
>>> import pandas as pd, cudf
>>> s = cudf.Series(
...     pd.date_range(start='2000-09-26', end='2000-10-03', freq='1D'))
>>> s
0   2000-09-26
1   2000-09-27
2   2000-09-28
3   2000-09-29
4   2000-09-30
5   2000-10-01
6   2000-10-02
7   2000-10-03
dtype: datetime64[ns]
>>> s.dt.is_quarter_start
0    False
1    False
2    False
3    False
4    False
5     True
6    False
7    False
dtype: bool
```

--------------------------------

### Import cuDF and NumPy for missing data examples

Source: https://docs.nvidia.com/cudf/latest/cudf/missing-data

Sets up the environment for the missing data examples by importing cudf and numpy, and creating a random number generator.

```python
import cudf
import numpy as np

rng = np.random.default_rng()

```

--------------------------------

### Install cudf nightly wheels

Source: https://docs.nvidia.com/cudf/latest/cudf_pandas/benchmarks

Install the required cudf nightly wheels based on the detected CUDA version.

```bash
CUDA_MAJOR=$(nvidia-smi | grep -oP 'CUDA Version: \K[0-9]+')
pip install --extra-index-url https://pypi.anaconda.org/rapidsai-wheels-nightly/simple/ \
    "cudf-cu${CUDA_MAJOR}>=0.0.0a0"
```

--------------------------------

### StringMethods.insert(start, repl)

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.insert

Inserts the specified string into each string in the Series or Index at the given position. The start parameter defines the beginning position for insertion, and repl is the string to insert. Returns a new Series or Index with the inserted string.

```APIDOC
## StringMethods.insert(start, repl)

### Description
Inserts the specified string into each string in the Series or Index at the given position. The start parameter defines the beginning position for insertion, and repl is the string to insert. Returns a new Series or Index with the inserted string.

### Method
insert

### Signature
StringMethods.insert(start: int = 0, repl: str | None = None) -> Series | Index

### Parameters
- **start** (int) - Optional - Beginning position of the string to replace. Default is beginning of each string. Specify -1 to insert at the end of each string.
- **repl** (str) - Optional - String to insert into the specified position value.

### Returns
- Series/Index of str dtype - A new string series with the specified string inserted at the specified position.

### Examples
```python
>>> import cudf
>>> s = cudf.Series(["abcdefghij", "0123456789"])
>>> s.str.insert(2, '_')
0    ab_cdefghij
1    01_23456789
dtype: str
```

When no repl is passed, nothing is inserted.
```python
>>> s.str.insert(2)
0    abcdefghij
1    0123456789
dtype: str
```

Negative values are also supported for start.
```python
>>> s.str.insert(-1,'_')
0    abcdefghij_
1    0123456789_
dtype: str
```
```

--------------------------------

### Install cudf

Source: https://docs.nvidia.com/cudf/latest/cudf_pandas/benchmarks

Install the cudf package for CUDA 12.

```bash
pip install cudf-cu12
```

--------------------------------

### Setup MultiIndex DataFrame

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.swaplevel/index.html

Initialize a DataFrame with a MultiIndex to demonstrate level swapping.

```python
>>> import cudf
>>> midx = cudf.MultiIndex(levels=[['llama', 'cow', 'falcon'],
...   ['speed', 'weight', 'length'],['first','second']],
...   codes=[[0, 0, 0, 1, 1, 1, 2, 2, 2], [0, 1, 2, 0, 1, 2, 0, 1, 2],
...             [0, 0, 0, 0, 0, 0, 1, 1, 1]])
>>> cdf = cudf.DataFrame(index=midx, columns=['big', 'small'],
...  data=[[45, 30], [200, 100], [1.5, 1], [30, 20],
...         [250, 150], [1.5, 0.8], [320, 250], [1, 0.8], [0.3, 0.2]])
```

--------------------------------

### Setup DataFrame for unstacking

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.unstack/index.html

Create a DataFrame with a MultiIndex to demonstrate unstacking operations.

```python
>>> df = cudf.DataFrame()
>>> df['a'] = [1, 1, 1, 2, 2]
>>> df['b'] = [1, 2, 3, 1, 2]
>>> df['c'] = [5, 6, 7, 8, 9]
>>> df['d'] = ['a', 'b', 'a', 'd', 'e']
>>> df = df.set_index(['a', 'b', 'd'])
>>> df
       c
a b d
1 1 a  5
  2 b  6
  3 a  7
2 1 d  8
  2 e  9
```

--------------------------------

### Initialize cuDF Series

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.match

Setup a Series of strings for matching operations.

```python
>>> import cudf
>>> s = cudf.Series(["rapids", "ai", "cudf"])
```

--------------------------------

### DataFrame division setup

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.div/index.html

Initialize a DataFrame for division operations.

```python
>>> df = cudf.DataFrame(
...     {'angles': [0, 3, 4], 'degrees': [360, 180, 360]},
...     index=['circle', 'triangle', 'rectangle']
... )
```

--------------------------------

### DataFrame division example

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.divide/index.html

Demonstrates creating a DataFrame and performing division by a scalar.

```python
>>> df = cudf.DataFrame(
...     {'angles': [0, 3, 4], 'degrees': [360, 180, 360]},
...     index=['circle', 'triangle', 'rectangle']
... )
```

```python
>>> df.truediv(1)
        angles  degrees
circle        0.0    360.0
triangle      3.0    180.0
rectangle     4.0    360.0
```

--------------------------------

### Index argsort example

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.argsort/index.html

Shows how to call argsort on a cudf.Index.

```python
>>> import cudf
>>> idx = cudf.Index([3, 1, 2])
>>> idx.argsort()
array([1, 2, 0])
```

--------------------------------

### Series division setup

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.div/index.html

Initialize two Series with missing values for division.

```python
>>> a = cudf.Series([1, 1, 1, None], index=['a', 'b', 'c', 'd'])
>>> b = cudf.Series([1, None, 1, None], index=['a', 'b', 'd', 'e'])
```

--------------------------------

### Initialize a Series

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.squeeze/index.html

Create a sample Series for demonstration.

```python
>>> primes = cudf.Series([2, 3, 5, 7])
```

--------------------------------

### rindex with start and end parameters

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.rindex

This example shows how to restrict the search range for the substring using the start and end parameters.

```python
>>> s = cudf.Series(['abc', 'abb','ab' ,'ddb'])
>>> s.str.rindex('b', start=1, end=5)
0    1
1    2
2    1
3    2
dtype: int64
```

--------------------------------

### Building and serving documentation

Source: https://docs.nvidia.com/cudf/latest/cudf/developer_guide/documentation

Commands to build the documentation using make and serve the generated HTML files locally using a Python HTTP server.

```bash
make html
```

```bash
python -m http.server
```

--------------------------------

### Initialize Series for split examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.split

Create a sample Series containing strings and a null value to demonstrate splitting operations.

```python
>>> import cudf
>>> data = ["this is a regular sentence",
...     "https://docs.python.org/index.html", None]
>>> s = cudf.Series(data)
>>> s
0            this is a regular sentence
1    https://docs.python.org/index.html
2                                   NaN
dtype: str
```

--------------------------------

### Get UTF-8 code points for string series

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.code_points

Examples showing how to retrieve UTF-8 code point values for characters in a cudf Series.

```python
>>> import cudf
>>> s = cudf.Series(["a","xyz", "éee"])
>>> s.str.code_points()
0       97
1      120
2      121
3      122
4    50089
5      101
6      101
dtype: int32
>>> s = cudf.Series(["abc"])
>>> s.str.code_points()
0    97
1    98
2    99
dtype: int32
```

--------------------------------

### Initialize DataFrame for set_index examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.set_index/index.html

Creates a sample DataFrame with columns 'a', 'b', and 'c' to be used in subsequent indexing operations.

```python
>>> df = cudf.DataFrame({
...     "a": [1, 2, 3, 4, 5],
...     "b": ["a", "b", "c", "d","e"],
...     "c": [1.0, 2.0, 3.0, 4.0, 5.0]
... })
>>> df
   a  b    c
0  1  a  1.0
1  2  b  2.0
2  3  c  3.0
3  4  d  4.0
4  5  e  5.0
```

--------------------------------

### Create a DataFrame for nsmallest examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.nsmallest/index.html

Initializes a sample DataFrame with population, GDP, and alpha-2 columns to demonstrate sorting operations.

```python
>>> import cudf
>>> df = cudf.DataFrame({'population': [59000000, 65000000, 434000,
...                                   434000, 434000, 337000, 337000,
...                                   11300, 11300],
...                    'GDP': [1937894, 2583560 , 12011, 4520, 12128,
...                            17036, 182, 38, 311],
...                    'alpha-2': ["IT", "FR", "MT", "MV", "BN",
...                                "IS", "NR", "TV", "AI"]},
...                   index=["Italy", "France", "Malta",
...                          "Maldives", "Brunei", "Iceland",
...                          "Nauru", "Tuvalu", "Anguilla"])
>>> df
          population      GDP alpha-2
Italy       59000000  1937894      IT
France      65000000  2583560      FR
Malta         434000    12011      MT
Maldives      434000     4520      MV
Brunei        434000    12128      BN
Iceland       337000    17036      IS
Nauru         337000      182      NR
Tuvalu         11300       38      TV
Anguilla       11300      311      AI
```

--------------------------------

### Extracting substrings with slice_from

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.slice_from

This example demonstrates how to use slice_from with two Series objects defining the start and stop indices for each string in the target Series.

```python
>>> import cudf
>>> s = cudf.Series(["hello","there"])
>>> s
0    hello
1    there
dtype: str
>>> starts = cudf.Series([1, 3])
>>> stops = cudf.Series([5, 5])
>>> s.str.slice_from(starts, stops)
0    ello
1      re
dtype: str
```

--------------------------------

### Initial Setup for cuDF UDFs

Source: https://docs.nvidia.com/cudf/latest/cudf/guide-to-udfs

Basic imports and configuration to suppress CUDA low occupancy warnings.

```python
import cudf
import numpy as np
from cudf.datasets import randomdata
from numba import config

config.CUDA_LOW_OCCUPANCY_WARNINGS = 0
```

--------------------------------

### Initialize DataFrame for isin examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.isin/index.html

Sets up a sample DataFrame with 'num_legs' and 'num_wings' columns for subsequent isin operations.

```python
>>> import cudf
>>> df = cudf.DataFrame({'num_legs': [2, 4], 'num_wings': [2, 0]},
...                     index=['falcon', 'dog'])
>>> df
        num_legs  num_wings
falcon         2          2
dog            4          0
```

--------------------------------

### Get days in month for a datetime Series

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.series.DatetimeProperties.days_in_month

This example demonstrates how to create a datetime Series using pandas date_range and then retrieve the number of days in the month for each entry using the .dt.days_in_month accessor.

```python
>>> import pandas as pd, cudf
>>> s = cudf.Series(
...     pd.date_range(start='2000-08-01', end='2001-08-01', freq='1M'))
>>> s
0    2000-08-31
1    2000-09-30
2    2000-10-31
3    2000-11-30
4    2000-12-31
5    2001-01-31
6    2001-02-28
7    2001-03-31
8    2001-04-30
9    2001-05-31
10   2001-06-30
11   2001-07-31
dtype: datetime64[ns]
>>> s.dt.days_in_month
0     31
1     30
2     31
3     30
4     31
5     31
6     28
7     31
8     30
9     31
10    30
11    31
dtype: int16
```

--------------------------------

### Describe all available options

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/options

Use cudf.describe_option() without arguments to print descriptions for all available configuration options.

```python
In [1]: import cudf

In [2]: cudf.describe_option()
default_integer_bitwidth:
	
Default bitwidth when the dtype of an integer needs to be
inferred. If set to `None`, the API will align dtype with pandas.
APIs that respect this option include:
	- cudf object constructors
	- cudf.read_csv and cudf.read_json when `dtype` is not specified.
	- APIs that require implicit conversion of cudf.RangeIndex to an
	  integer index.
	Valid values are  None, 32 or 64. Default is None.

	[Default: None] [Current: None]
default_float_bitwidth:
	
Default bitwidth when the dtype of a float needs to be
inferred. If set to `None`, the API will align dtype with pandas.
APIs that respect this option include:
	- cudf object constructors
	- cudf.read_csv and cudf.read_json when `dtype` is not specified.
	Valid values are None, 32 or 64. Default is None.

	[Default: None] [Current: None]
spill:
	
Enables spilling.
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
spill_on_demand:
	
Enables spilling on demand using an RMM out-of-memory error handler.
This has no effect if spilling is disabled, see the "spill" option.
	Valid values are True or False. Default is True.

	[Default: True] [Current: True]
spill_device_limit:
	
Enforce a device memory limit in bytes.
This has no effect if spilling is disabled, see the "spill" option.
	Valid values are any positive integer or None (disabled).
	Default is None.

	[Default: None] [Current: None]
spill_stats:
	
If not 0, enables statistics at the specified level:
    0  - disabled (no overhead).
    1+ - duration and number of bytes spilled (very low overhead).
    2+ - a traceback for each time a spillable buffer is exposed
        permanently (potential high overhead).

Valid values are any positive integer.
Default is 0 (disabled).

	[Default: 0] [Current: 0]
mode.pandas_compatible:
	
If set to `False`, retains `cudf` specific behavior.
If set to `True`, enables pandas compatibility mode,
which will try to match pandas API behaviors in case of
any inconsistency.
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
memory_profiling:
	
If set to `False`, disables memory profiling.
If set to `True`, enables memory profiling.
Read more at: :ref:`memory-profiling-user-doc`
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
io.parquet.low_memory:
	
If set to `False`, reads entire parquet in one go.
If set to `True`, reads parquet file in chunks.
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
io.json.low_memory:
	
If set to `False`, reads entire json in one go.
If set to `True`, reads json file in chunks.
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
kvikio_remote_io:
	
Whether to use KvikIO's remote IO backend or not.
	WARN: this is experimental and may be removed at any time
	without warning or deprecation period.
	Set KVIKIO_NTHREADS (default is 8) to change the number of
	concurrent tcp connections, which is important for good performance.
	Valid values are True or False. Default is False.

	[Default: False] [Current: False]
```

--------------------------------

### Numpydoc style docstring example

Source: https://docs.nvidia.com/cudf/latest/cudf/developer_guide/documentation

An example of a class and method docstring following the numpydoc style, including parameters, return values, exceptions, and examples.

```python
class A:
    """Brief description of A.

    Longer description of A.

    Parameters
    ----------
    x : int
        Description of x, the first constructor parameter.
    """
    def __init__(self, x: int):
        pass

    def foo(self, bar: str):
        """Short description of foo.

        Longer description of foo.

        Parameters
        ----------
        bar : str
            Description of bar.

        Returns
        -------
        float
            Description of the return value of foo.

        Raises
        ------
        ValueError
            Explanation of when a ValueError is raised.
            In this case, a ValueError is raised if bar is "fail".

        Examples
        --------
        The examples section is _strongly_ encouraged.
        Where appropriate, it may mimic the examples for the corresponding pandas API.
        >>> a = A()
        >>> a.foo('baz')
        0.0
        >>> a.foo('fail')
        ...
        ValueError: Failed!
        """
        if bar == "fail":
            raise ValueError("Failed!")
        return 0.0
```

--------------------------------

### Initialize DataFrame for renaming examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.rename/index.html

Create a sample DataFrame with columns 'A' and 'B' to demonstrate renaming operations.

```python
>>> import cudf
>>> df = cudf.DataFrame({"A": [1, 2, 3], "B": [4, 5, 6]})
>>> df
   A  B
0  1  4
1  2  5
2  3  6
```

--------------------------------

### Create a Series for head() examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.head/index.html

Initializes a cuDF Series with string data for demonstration.

```python
>>> ser = cudf.Series(['alligator', 'bee', 'falcon',
... 'lion', 'monkey', 'parrot', 'shark', 'whale', 'zebra'])
>>> ser
0    alligator
1          bee
2       falcon
3         lion
4       monkey
5        parrot
6        shark
7        whale
8        zebra
dtype: str
```

--------------------------------

### Match strings starting with 'a'

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.match

Check if each string in the Series starts with the character 'a'.

```python
>>> s.str.match('a')
0    False
1     True
2    False
dtype: bool
```

--------------------------------

### Install cudf-polars with conda

Source: https://docs.nvidia.com/cudf/latest/cudf_polars

Install the cudf-polars package using conda from the rapidsai, conda-forge, and nvidia channels.

```bash
conda install -c rapidsai -c conda-forge -c nvidia cudf-polars
```

--------------------------------

### Import libraries and check version

Source: https://docs.nvidia.com/cudf/latest/cudf/performance-comparisons/performance-comparisons

Initializes the environment by importing necessary libraries and printing the cuDF version.

```python
import timeit

import cudf
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

print(f"{cudf.__version__=}")
```

--------------------------------

### Initialize Series for findall examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.findall

Sets up a cudf Series containing strings to be used in subsequent findall operations.

```python
>>> import cudf
>>> s = cudf.Series(['Lion', 'Monkey', 'Rabbit'])
```

--------------------------------

### Replace from start index to end

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.slice_replace

Specify only the start index to replace the substring from that position until the end of each string.

```python
>>> s.str.slice_replace(1, repl='X')
0    aX
1    aX
2    aX
3    aX
4    aX
dtype: str
```

--------------------------------

### Initialize a DataFrame for rounding

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.round/index.html

Sets up a sample DataFrame with two columns to demonstrate rounding operations.

```python
>>> df = cudf.DataFrame(
...     [(.21, .32), (.01, .67), (.66, .03), (.21, .18)],
...     columns=['dogs', 'cats'],
... )
>>> df
   dogs  cats
0  0.21  0.32
1  0.01  0.67
2  0.66  0.03
3  0.21  0.18
```

--------------------------------

### Find substring with start and end parameters

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.find

Shows how to restrict the search range using the start and end parameters.

```python
>>> s.str.find('b', start=1, end=5)
0    1
1   -1
2   -1
3    2
dtype: int64
```

--------------------------------

### Initialize Series for replacement examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.replace

Sets up a cuDF Series containing string data and a null value to demonstrate replacement operations.

```python
>>> import cudf
>>> s = cudf.Series(['foo', 'fuz', None])
>>> s
0     foo
1     fuz
2    NaN
dtype: str
```

--------------------------------

### StringMethods.slice_from(starts, stops)

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.slice_from

Returns a substring of each string using positions for each string. The starts and stops parameters are of Series type.

```APIDOC
## StringMethods.slice_from(starts, stops)

### Description
Return substring of each string using positions for each string. The starts and stops parameters are of Series type.

### Method
slice_from

### Signature
StringMethods.slice_from(starts: Series, stops: Series) -> Series | Index

### Parameters
- **starts** (Series) - Required - Beginning position of each string to extract. Default is beginning of each string.
- **stops** (Series) - Required - Ending position of each string to extract. Default is end of each string. Use -1 to specify to the end of that string.

### Returns
- **Series/Index of str dtype** - A substring of each string using positions for each string.

### Example
```python
>>> import cudf
>>> s = cudf.Series(["hello","there"])
>>> s
0    hello
1    there
dtype: str
>>> starts = cudf.Series([1, 3])
>>> stops = cudf.Series([5, 5])
>>> s.str.slice_from(starts, stops)
0    ello
1      re
dtype: str
```
```

--------------------------------

### Using start and end parameters with rfind

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.rfind

Shows how to constrain the search range for the substring using the start and end parameters.

```python
>>> s.str.rfind('a', start=2, end=5)
0   -1
1   -1
2   -1
dtype: int64
```

--------------------------------

### Match strings starting with 'a' or 'c'

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.match

Check if each string in the Series starts with either 'a' or 'c' using a regex character class.

```python
>>> s.str.match('[ac]')
0    False
1     True
2     True
dtype: bool
```

--------------------------------

### Using StringMethods.center to pad strings

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.core.accessors.string.StringMethods.center

Examples showing how to center strings in a cuDF Series with different widths and fill characters.

```python
>>> import cudf
>>> s = cudf.Series(['a', 'b', None, 'd'])
>>> s.str.center(1)
0       a
1       b
2    NaN
3       d
dtype: str
>>> s.str.center(1, fillchar='-')
0       a
1       b
2    NaN
3       d
dtype: str
>>> s.str.center(2, fillchar='-')
0      a-
1      b-
2    NaN
3      d-
dtype: str
>>> s.str.center(5, fillchar='-')
0    --a--
1    --b--
2     NaN
3    --d--
dtype: str
>>> s.str.center(6, fillchar='-')
0    --a---
1    --b---
2      NaN
3    --d---
dtype: str
```

--------------------------------

### Unsupported UDF operation example

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.apply/index.html

This example demonstrates a UDF that triggers an AttributeError because it attempts to use the .upper() method on a string column, which is not supported.

```python
>>> def f(row):
...     st = row['str_col'].upper()
...     return 'ABC' in st
>>> df.apply(f, axis=1)
```

--------------------------------

### Series take example

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.take/index.html

Demonstrates selecting rows from a cudf.Series using a list of indices.

```python
>>> s = cudf.Series(['a', 'b', 'c', 'd', 'e'])
>>> s.take([2, 0, 4, 3])
2 c
0 a
4 e
3 d
dtype: str
```

--------------------------------

### Install cudf-polars with pip

Source: https://docs.nvidia.com/cudf/latest/cudf_polars

Install the cudf-polars package using pip. Use the cu13 suffix for CUDA 13 or cu12 for CUDA 12.

```bash
pip install cudf-polars-cu13
```

--------------------------------

### Create virtual environment for pandas

Source: https://docs.nvidia.com/cudf/latest/cudf_pandas/benchmarks

Create a virtual environment for the pandas benchmark dependencies.

```bash
virtualenv pandas/py-pandas
```

--------------------------------

### Create a cuDF DataFrame for iloc examples

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.iloc/index.html

Creates a DataFrame with columns a, b, and c, each containing values from 0 to 19. This DataFrame is used in the subsequent iloc selection examples.

```python
>>> df = cudf.DataFrame({'a': range(20),
...                      'b': range(20),
...                      'c': range(20)})
```

--------------------------------

### Initialize DataFrame for conversion

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.convert_dtypes/index.html

Create a DataFrame with various dtypes to demonstrate the conversion process.

```python
>>> import cudf
>>> import numpy as np
>>> df = cudf.DataFrame(
...     {
...         "a": cudf.Series([1, 2, 3], dtype=np.dtype("int32")),
...         "b": cudf.Series(["x", "y", "z"], dtype=np.dtype("O")),
...         "e": cudf.Series([10, np.nan, 20], dtype=np.dtype("float")),
...         "f": cudf.Series([np.nan, 100.5, 200], dtype=np.dtype("float")),
...     }
... )
```

--------------------------------

### DataFrame.keys()

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/dataframe

Get the columns.

```APIDOC
## DataFrame.keys()

### Description
Get the columns.

### Method
keys

### Returns
Index of column labels.
```

--------------------------------

### Series.name

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/series

Get the name of this object.

```APIDOC
## Series.name

### Description
Get the name of this object.
```

--------------------------------

### Build DataFrame with __setitem__

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame/index.html

Demonstrates creating an empty DataFrame and adding columns using the __setitem__ syntax.

```python
>>> import cudf
>>> df = cudf.DataFrame()
>>> df['key'] = [0, 1, 2, 3, 4]
>>> df['val'] = [float(i + 10) for i in range(5)]  # insert column
>>> df
   key   val
0    0  10.0
1    1  11.0
2    2  12.0
3    3  13.0
4    4  14.0
```

--------------------------------

### Series.index

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/series

Get the labels for the rows.

```APIDOC
## Series.index

### Description
Get the labels for the rows.
```

--------------------------------

### Sample with replacement from a Series

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/api/cudf.DataFrame.sample/index.html

Demonstrates sampling 10 items from a Series with replacement enabled.

```python
>>> sr = cudf.Series([1, 2, 3, 4, 5])
>>> sr.sample(10, replace=True, random_state=0)
1    2
4    5
1    2
2    3
4    5
1    2
4    5
2    3
4    5
1    2
dtype: int64
```

--------------------------------

### Series.dtype

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/series

Get the data type of the Series.

```APIDOC
## Series.dtype

### Description
Get the data type of the Series.
```

--------------------------------

### cudf.get_option

Source: https://docs.nvidia.com/cudf/latest/cudf/api_docs/options

Get the value of the specified option.

```APIDOC
## cudf.get_option(name)

### Description
Get the value of the specified option.

### Parameters
- **name** (str) - Required - The name of the option to retrieve.

### Returns
The current value of the option.
```
Tokens:
Raw