Metadata-Version: 2.4
Name: db-robust-clust
Version: 0.1.8
Summary: Apply distance based robust clustering for mixed data.
Home-page: https://github.com/FabioScielzoOrtiz/db_robust_clust
Author: Fabio Scielzo Ortiz
Author-email: fabio.scielzoortiz@gmail.com
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.7
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: polars
Requires-Dist: numpy<=1.26.4
Requires-Dist: robust-mixed-dist
Requires-Dist: pandas
Requires-Dist: scikit-learn-extra
Requires-Dist: tqdm
Requires-Dist: setuptools
Requires-Dist: pyarrow
Requires-Dist: matplotlib
Requires-Dist: seaborn
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: license-file
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# `db-robust-clust`

In the era of big data, data scientists are trying to solve real-world problems using multivariate
and heterogeneous datasets, i.e., datasets where for each unit multiple variables of different
nature are observed. Clustering may be a challenging problem when data are of mixed-type and
present an underlying correlation structure and outlying units.

In the paper ***Grané, A., Scielzo-Ortiz, F.: New distance-based clustering algorithms for large mixed-type data, Submitted to Journal of Classification (2025)***, new efficient robust clustering algorithms able to deal with large mixed-type data are developed and implemented in a **new Python package**, called `db-robust-clust`, hosted in the official Python Package Index (PyPI), the standard repository of packages for the Python programming language:: https://pypi.org/project/db_robust_clust/. 

Their performance is analyzed in rather complex mixed-type datasets,
both synthetic and real, where a wide variety of scenarios is considered regarding
size, the proportion of outlying units, the underlying correlation structure, and the
cluster pattern. The simulation study comprises four computational experiments
conducted on datasets of sizes ranging from 35k to 1M, in which the accuracy and
efficiency of the new proposals are tested and compared to those of existing clus-
tering alternatives. In addition, the goodness and computing time of the methods
under evaluation are tested on real datasets of varying sizes and patterns. MDS is
used to visualize clustering results.
