HashSlap-Summer-of-Code/ml-core

Standardize Dataset Handling Across All ML Modules

Aperta

#11 aperta il 18 giu 2025

 (0 commenti) (0 reazioni) (0 assegnatari)Jupyter Notebook (11 fork)auto 404
Intermediateenhancementgood first issuehacktoberfesthssoc

Metriche repository

Star
 (3 stelle)
Metriche merge PR
 (Metriche PR in attesa)

Descrizione

Description:
Right now, different subfolders load and process datasets in inconsistent ways. Create a unified, reusable Python module to handle dataset loading and basic preprocessing (e.g., scaling, splitting). This ensures maintainability and reduces repeated code across ML scripts.

Expected Tasks:

  • Create a Python utility (e.g., data_utils.py) with functions like:
    • load_csv(path)
    • train_test_split(X, y, test_size=0.2)
    • standardize(X)
  • Save this file in a utils/ folder.
  • Refactor at least two existing implementations (e.g., KNN and Perceptron) to use this utility.
  • Update documentation in the root README.md and/or affected folders.

Stretch Goal:

  • Add optional support for downloading public datasets (e.g., from UCI or sklearn).

Guida contributor