[MENTEE] Shreesha M

[ShreeshaM07]

Mentee:
Shreesha M (he/him) , Linkedin

Mentor(s):
@fkiraly @kirilral @astrogilda

Why did you join sktime's mentorship program?
As a machine learning enthusiast I like to explore about the various algorithms present and different techniques available. Implementing some of these algorithms while contributing to sktime would enhance my knowledge about the various algorithms and their working and the various scenarios they can be used. Sktime being an open source organisation helps me develop these skills. Under supervision of my mentors I can achieve my goals effectively.

What topics are you working on?

• Interfacing of the NGBoost package in skpro in the PRs
  • Regressor :[ENH] interface to ngboost
  • Survival : [ENH] Interface NGBSurvival from ngboost
  • Refactor : [MNT] refactor NGBoostRegressor to inherit NGBoostAdapter
  • Coupling reduction : [ENH] Exponential dist in NGBoostRegressor, NGBoostSurvival
• Added the Exponential Distribution [ENH] Exponential Distribution
• Implemented Gamma distribution [ENH] Gamma Distribution
• Histogram distribution
  • Design for ragged array distributions [ENH] Histogram distribution
  • Implemented Histogram Distribution and also created base class for ragged array distributions [ENH] Histogram distribution
  • Handled bug caused due to deprecation in numpy v2.0.0 [BUG] Histogram Distribution: address np.broadcast_arrays deprecation of writable return in numpy 2.0.0
• Bug fix in TestAllDistributions in skpro specifically in [BUG] fix test_methods_p logic when shuffle is True
• Generalized Linear Models : Implemented support for multiple families/distributions and multiple link functions.Designed a way to pass offset and exposure parameters. [ENH] GLM with multiple distributions and link function support
• Reduction to classification
  • Implemented a regressor which bins the y into classes and then fits a classifier to predict probabilities of each class and fits it to a Histogram Distribution [ENH] Multiclass classification reduction using Histograms

What are your learning goals?

• Gain a deeper understanding of the working/methodology/math of some ML algorithms.
• Understand how to pick the right algorithms in different scenarios.
• Familiarize with AI tools and methods and get practical ML experience.

What's next for you after the mentorship program?

I would love to continue contributing for sktime and work on other interesting projects.