"""Load data for morphing."""
from __future__ import annotations
from importlib.resources import files
from itertools import zip_longest
from numbers import Number
from pathlib import Path
from typing import ClassVar
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from matplotlib.axes import Axes
from .. import MAIN_DIR
from ..plotting.style import plot_with_custom_style
from .dataset import Dataset
[docs]
class DataLoader:
"""
Class for loading datasets for morphing.
.. plot::
:caption:
Datasets currently included in Data Morph. The ``dino``
shape is Alberto Cairo's `Datasaurus`_. All logos are used
with permission (see below). Note that CSV files are also supported
by the :meth:`.load_dataset` method.
from data_morph.data.loader import DataLoader
DataLoader.plot_available_datasets()
.. _Datasaurus: https://web.archive.org/web/20240620205540/http://www.thefunctionalart.com/2016/08/download-datasaurus-never-trust-summary.html
.. important:: All logos are used with permission.
* The Python logo is a `trademark of the Python Software Foundation (PSF)
<https://www.python.org/psf/trademarks/>`_, used with permission from
the Foundation.
* The `SuperDataScience (SDS) <https://www.superdatascience.com/podcast>`_
logo is used with permission.
"""
_DATA_PATH: str = 'data/starter_shapes/'
_DATASETS: ClassVar[dict[str, str]] = {
'bunny': 'bunny.csv',
'cat': 'cat.csv',
'dino': 'dino.csv',
'dog': 'dog.csv',
'gorilla': 'gorilla.csv',
'music': 'music.csv',
'panda': 'panda.csv',
'pi': 'pi.csv',
'Python': 'python.csv',
'SDS': 'superdatascience.csv',
'sheep': 'sheep.csv',
'soccer': 'soccer_ball.csv',
}
AVAILABLE_DATASETS = sorted(_DATASETS.keys(), key=str.lower)
"""list[str]: List of available built-in starter datasets,
which can be visualized with :meth:`plot_available_datasets`."""
def __init__(self) -> None:
raise NotImplementedError
[docs]
@classmethod
def load_dataset(
cls,
dataset: str,
scale: Number | None = None,
) -> Dataset:
"""
Load dataset.
Parameters
----------
dataset : str
Either one of :attr:`AVAILABLE_DATASETS` or a path to a
CSV file containing two columns: x and y.
scale : numbers.Number, optional
The factor to scale the data by (can be used to speed up morphing).
Values in the data's x and y columns will be divided by this value.
Returns
-------
Dataset
The starting dataset for morphing.
Notes
-----
If you are looking to create a :class:`.Dataset` from a
:class:`~pandas.DataFrame` object, use the :class:`.Dataset`
class directly.
"""
if dataset == 'python':
dataset = dataset.capitalize()
elif dataset == 'sds':
dataset = dataset.upper()
try:
filepath = files(MAIN_DIR).joinpath(
Path(cls._DATA_PATH) / cls._DATASETS[dataset]
)
name = dataset
df = pd.read_csv(filepath)
except KeyError:
try:
name = Path(dataset).stem
df = pd.read_csv(dataset)
except FileNotFoundError as err:
raise ValueError(
f'Unknown dataset "{dataset}". '
'Provide a valid path to a CSV dataset or use one of '
f'the included datasets: {", ".join(cls.AVAILABLE_DATASETS)}.'
) from err
return Dataset(name=name, df=df, scale=scale)
[docs]
@classmethod
@plot_with_custom_style
def plot_available_datasets(cls) -> Axes:
"""
Plot the built-in datasets.
Returns
-------
matplotlib.axes.Axes
The :class:`~matplotlib.axes.Axes` object containing the plot.
See Also
--------
AVAILABLE_DATASETS
The list of available datasets built into Data Morph.
"""
num_plots = len(cls.AVAILABLE_DATASETS)
num_cols = 3
num_rows = int(np.ceil(num_plots / num_cols))
fig, axs = plt.subplots(
num_rows,
num_cols,
layout='constrained',
figsize=(12, 4 * num_rows),
subplot_kw={'aspect': 'equal'},
)
fig.get_layout_engine().set(w_pad=0.2, h_pad=0.2)
for dataset, ax in zip_longest(cls.AVAILABLE_DATASETS, axs.flatten()):
if dataset:
ax.tick_params(
axis='both',
which='both',
bottom=False,
left=False,
right=False,
labelbottom=False,
labelleft=False,
)
points = cls.load_dataset(dataset)
if dataset == 'Python':
dataset += ' logo\u2122'
elif dataset == 'SDS':
dataset += ' logo'
ax.scatter(points.df.x, points.df.y, s=4, color='black')
# tight plot bounds for the grid of datasets in the docs
bounds = points.data_bounds.clone()
x_offset, y_offset = (offset * 0.1 for offset in bounds.range)
bounds.adjust_bounds(x=x_offset, y=y_offset)
bounds.align_aspect_ratio()
ax.set(
title=f'{dataset} ({points.df.shape[0]:,d} points)',
xlim=bounds.x_bounds,
ylim=bounds.y_bounds,
xlabel='',
ylabel='',
)
else:
ax.remove()
return axs
