"""
Generate synthetic high-dimensional classification datasets.
This module creates multiple configurations of multi-class classification datasets
with varying dimensionality, feature characteristics, and class distributions,
useful for testing machine learning algorithms on high-dimensional data.
"""
import itertools
import json
import os
import numpy as np
import pandas as pd
from sklearn.datasets import make_classification
dataset_config = {}
# parameters to vary across the configurations
N_SAMPLES = list(range(100, 300, 50))
N_FEATURES = list(range(10, 60, 10))
N_INFORMATIVE = list(range(2, 8, 4))
N_REDUNDANT = list(range(2, 8, 4))
N_CLASSES = list(range(2, 4, 6))
N_CLUSTERS_PER_CLASS = list(range(1, 2, 3))
WEIGHTS = [[0.3, 0.7], [0.4, 0.6], [0.5, 0.5]]
[docs]
def generate_classification_datasets(
n_samples,
n_features,
n_informative,
n_redundant,
n_classes,
n_clusters_per_class,
weights,
save_path=None,
random_state=42,
):
"""
Generate multiple high-dimensional classification datasets with varying parameters.
Creates a series of synthetic datasets for multi-class classification problems
with configurable feature characteristics including informative features,
redundant features, and class distributions.
Parameters
----------
n_samples : list of int
List of sample sizes to generate for each configuration.
n_features : list of int
List of total feature counts (must be >= n_informative + n_redundant).
n_informative : list of int
List of informative feature counts that are useful for prediction.
n_redundant : list of int
List of redundant feature counts (linear combinations of informative features).
n_classes : list of int
List of class counts for multi-class classification.
n_clusters_per_class : list of int
List of cluster counts per class.
weights : list of list of float
List of class weight distributions (must sum to 1.0).
save_path : str, optional
Directory path where datasets and configuration files will be saved.
random_state : int, default=42
Random seed for reproducibility.
Returns
-------
None
Saves CSV files for each dataset configuration and a JSON file with
all configuration parameters.
Notes
-----
- Each dataset is saved as 'class_data-{i}.csv' where i is the configuration number
- Configuration parameters are saved in 'dataset_config.json'
- The last column 'class' contains class labels
- Only valid configurations where (n_informative + n_redundant) <= n_features are generated
Examples
--------
>>> from qbiocode.data_generation import generate_classification_datasets
>>> generate_classification_datasets(
... n_samples=[100], n_features=[20], n_informative=[5],
... n_redundant=[2], n_classes=[2], n_clusters_per_class=[1],
... weights=[[0.5, 0.5]], save_path='data'
... )
Generating classes dataset...
"""
print("Generating classes dataset...")
np.random.seed(random_state)
if save_path is None:
save_path = "class_data"
if not os.path.exists(save_path):
os.makedirs(save_path)
# enumerate all possible combinations of parameters based on ranges above
configurations = list(
itertools.product(
*[
n_samples,
n_features,
n_informative,
n_redundant,
n_classes,
n_clusters_per_class,
weights,
]
)
)
count_configs = 1
# populate all the configs with the corresponding argument values
for n_s, n_f, n_i, n_r, n_cla, n_clu, weights in configurations:
if (n_i + n_r) <= n_f:
config = "n_samples={}, n_features={}, n_informative={}, n_redundant={}, n_classes={}, n_clusters_per_class={}, weights={}".format(
n_s, n_f, n_i, n_r, n_cla, n_clu, weights
)
# print(count_configs)
# iteratively run the function for each combination of arguments
X, y = make_classification(
n_samples=n_s,
n_features=n_f,
n_informative=n_i,
n_redundant=n_r,
n_classes=n_cla,
n_clusters_per_class=n_clu,
weights=weights,
random_state=random_state,
)
# print("Configuration {}/{}: {}".format(count_configs, len(configurations), config))
dataset = pd.DataFrame(X)
dataset["class"] = y
with open(os.path.join(save_path, "dataset_config.json"), "w") as outfile:
dataset_config.update(
{
"hd_data-{}.csv".format(count_configs): {
"n_samples": n_s,
"n_features": n_f,
"n_informative": n_i,
"n_redundant": n_r,
"n_classes": n_cla,
"n_clusters_per_class": n_clu,
"weights": weights,
}
}
)
json.dump(dataset_config, outfile, indent=4)
new_dataset = dataset.to_csv(
os.path.join(save_path, "class_data-{}.csv".format(count_configs)), index=False
)
count_configs += 1
# print(X.shape)
# print(y.shape)
return
