Advestis is now fully integrated into Forvis Mazars, allowing us to operate as a unified team.
- TransparentPath : A pathlib-like way to handle paths on Google Cloud Storage. Developed by our former Lead Python Engineer Philippe Cotte.
- MailUtility : Package containing a tool to send mail and a tool to monitor a mailbox easily. Supports remote directories with TransparentPath.
- TableWriter : Class used to produce a ready-to-compile .tex file containing a table from a pandas or dask DataFrame object. Can also compile the .tex to produce a .pdf.
- PDFFactory : Class to update and/or create a PDF object, and add pages to it. Supports matplotlib.figure and tablewriter.TableWriter objects.
- HtmlMerger : Package allowing to merge all HTML files in a directory in a single file.
- Color and Style Cycler : Cycler for matplotlib. Can cycle through combination of colors and line or marker styles.
- AdRubix : RubixHeatmap class for plotting complex, highly customizable heatmaps with metadata as HTML and PNG.
- AdParallelEngine : A package wrapped around Dask, mpi4py and basic Python's multiprocessing libraries providing an easy way to make parallel maps.
- AdAdjust : Package allowing to fit any mathematical function to (for now 1D only) data.
- AdAnnealing : Package doing simulated annealing.
- IFRA : Interpretable Federated Rule Algorithm.
- RIPE (forked from thibaulthans/RIPE) : Implementation of a rule based prediction algorithm called RIPE.
- RICE : Implementation of a rule-based prediction algorithm called RICE (Rule Induction Covering Estimator). RICE is a deterministic and interpretable algorithm, for regression problem.
- ruleskit : Package implementing all useful tools for rule-based machine learning algorithms.
- els_paper : Repository for the Elsevier Energy and AI 2022 paper "Stress Testing Electrical Grids: Generative Adversarial Networks For Load Scenario Generation".
- ctgan : Conditional GAN for generating synthetic tabular data.
- Interpretability : Code and Data for the paper A rigorous method to compare interpretability of rule-based algorithms.
- adNMTF : Non-Negative Matrix and Tensor Factorizations.
- adILSM : ILSM is Integrated Latent Multi Source Model.
- nmf-rank-determination : Official GitHub repository for the original MDPI Mathematics paper.
- enAInem : Algorithms for decomposing nonnegative multiway array and multi-view data into rank-1. nonnegative tensors.
Note that enAInem combines adNMTF and adILSM, which are maintained for legacy purposes.
Our publications are available here : https://www.advestis.com/research.