This dataset contains the source code for uncertainty-aware principal component analysis (UA-PCA) and a series of images that show dimensionality reduction plots created with UA-PCA.

The software is a JavaScript library for performing principal component analysis and dimensionality reduction on datasets consisting of multivariate probability distributions.

Each plot of the image series used UA-PCA to project a dataset consisting of multivariate normal distributions. The covariance matrices of the dataset instances were scaled with different factors resulting in different UA-PCA projections. The projected probability distributions are displayed using isolines of their probability density functions. As the scaling value increases, the projection changes, showing the sensitivity of UA-PCA to changes in variance.

For build instructions and examples please refer to the README.md file in the file archive.

The dataset shown in the images is the 'student grades data set' that was also used in the related publication 'Uncertainty-Aware Principal Component Analysis' by Görtler et al. and originally published by Denoeux and Masson in their work 'Principal component analysis of fuzzy data using autoassociative neural networks'. It consists of grade descriptions for four different school subjects. The grade descriptions exhibit different levels of uncertainty of each student's performance.

Use persistent identifiers from Software Heritage (

) to cite individual files or even lines of the source code.