Built with pure Numpy and Matplotlib. Wanna make your own?
I'm Badr MOUFAD and I'm a Ph.D. candidate at Ecole Polytechnique–CMAP working on Denoising Diffusion Probabilistic Models and Markov Chain Monte Carlo methods for mobility trajectory generation. Before that, I was a Research Engineer at Inria Lyon, team OCKHAM. There, I developed fast and modular solvers for regularized GLMs.
In parallel, I am an open source enthusiast, I'm a maintainer of
skglm
, a fast and modular scikit-learn replacement for sparse GLMsBenchopt
, a collaborative tool to perform reproducible and transparent benchmarks between solversceler
, a state-of-the-art solver for Lasso Like problemsskillner
, a rule-based NLP package that automatically extracts skills from text
I have a major in Industrial management engineering and a minor in data science from EMINES – Mohammed VI Polytechnic University (UM6P).
I like to share my thoughts through simple, hands-on, and well-illustrated blogs. So far, I have written:
- Leveraging linear regression for feature selection of categorical and continuous variables, where I describe an approach to select top-K most relevant features that explains a target variable
- It’s time to structure your data science project, where I propose a structure to organize data science project to improve their scalability and reproducibility
- Deriving the inverse transform sampling method from first principles, where I build from scratch a tool to generate samples from a custom probability density function