Hi,
I have been studying the gamboostLSS framework and exploring its implementation of boosting-based distributional regression models, particularly how distribution families define likelihood and gradient functions within the boosting process.
As part of my exploration, I have been working on initial experiments related to extending the framework towards modeling dependencies between multiple response variables using copula-based approaches.
I am particularly interested in contributing towards integrating distributional copula regression into gamboostLSS. As a starting point, I would like to:
- further understand the design of existing family implementations,
- explore how a CopulaFamily() interface could be structured in alignment with the current architecture,
- and contribute through smaller steps such as improving documentation, adding examples, or prototyping components related to this extension.
I have been maintaining my exploratory work in a separate repository:
https://github.com/mankameshwarmishra5-cmd/gamboostLSS-project
This repository contains my initial experiments and exploration related to this idea.
I would appreciate any guidance on whether this direction aligns with the project’s goals, and if there are specific areas where I can start contributing effectively.
Thank you.
Hi,
I have been studying the gamboostLSS framework and exploring its implementation of boosting-based distributional regression models, particularly how distribution families define likelihood and gradient functions within the boosting process.
As part of my exploration, I have been working on initial experiments related to extending the framework towards modeling dependencies between multiple response variables using copula-based approaches.
I am particularly interested in contributing towards integrating distributional copula regression into gamboostLSS. As a starting point, I would like to:
I have been maintaining my exploratory work in a separate repository:
https://github.com/mankameshwarmishra5-cmd/gamboostLSS-project
This repository contains my initial experiments and exploration related to this idea.
I would appreciate any guidance on whether this direction aligns with the project’s goals, and if there are specific areas where I can start contributing effectively.
Thank you.