Reval

Latest version: v1.1.0

Safety actively analyzes 681775 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

1.1.2

Fixed the verbose option of FindBestClustCV.best_nclust() method, which now includes the validation performance metrics. Adde horizontal lines in the plot_mtrics() function that mark the performance of a random classifier.

1.1.1

Enable cluster ranges with step > 1 and verbose running of (repeated) cross-validation.

1.1

Added repeated cross validation to estimate clustering stability.

1.0.0

Determining the number of clusters that best partitions a dataset can be a challenging task because of 1) the lack of a priori information within an unsupervised learning framework; and 2) the absence of a unique clustering validation approach to evaluate clustering solutions. reval is a Python package that leverages stability-based relative clustering validation methods to determine best clustering solutions, as described in [1].

Statistical software, both in R and Python, usually compute internal validation metrics that can be leveraged to select the number of clusters that best fit the data and open-source software solutions that easily implement relative clustering techniques are lacking. The advantage of a relative approach over internal validation methods lies in the fact that internal metrics exploit characteristics of the data itself to produce a result, whereas relative validation converts an unsupervised clustering algorithm into a supervised classification problem, hence enabling generalizability and replicability of the results.

[1] Lange, T., Roth, V., Braun, M. L., & Buhmann, J. M. (2004). Stability-based validation of clustering solutions. Neural computation, 16(6), 1299-1323.

0.1.0

Added HDBSCAN to the clustering algorithms that can be selected; implemented hyperparameter selection (classifier/clustering combination and within algorithms); added parallel processing; speeded up computations.

Links

Releases

Has known vulnerabilities

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.