This project is motivated by our research on how to deal with high dimensional data, in which the number of variables in a dataset is larger than the number of observations. This high dimensional problem has become very common in recent years, and can be found in many areas like biomedicine, economics, climatology, etc. One of the best solutions when dealing with this type of data is using a penalization approach, amongst which the sparse group lasso, and its adaptive version are two of the best alternatives.
asgl (whose name stands from Adaptive Sparse Group Lasso) is a Python package (available at the Pypi repository) that solves penalized linear regression and quantile regression models for simultaneous variable selection and prediction, in both high and low dimensional frameworks. This package brings to Python many features that were already available in R packages (like solving sgl models) and it also provides many extra features not available in other programming languages, like the possibility of using adaptive models, known to achieve better results than the non-adaptive counterparts.