jay2323-tech/amazon-ml-challenge-2026

How jay2323-tech/amazon-ml-challenge-2026 works

This repository is a starter scaffold for a business entity-resolution challenge, not a complete matching application. The README defines the intended workflow—generate candidates, score them, and emit matches and candidates—but the blocking, feature, model, and submission packages contain no sampled implementation. The graph separates that documented workflow from the implemented normalization, evaluation, split-generation, EDA, and submission-validation utilities; links between unsampled pipeline stages are therefore documented rather than code-verified.

Main entry points: validate_submission.py Submission validator · make_split.py Validation split · blocking Candidate blocking · f05.py F0.5 scoring

Components 17

Matching workflow

Source records
Text normalization code/business_entity_resolution/src/preprocessing/normalize.py
Candidate blocking code/business_entity_resolution/src/blocking
Match features code/business_entity_resolution/src/features
Match model code/business_entity_resolution/src/models
Submission output code/business_entity_resolution/src/submission
Candidate pairs
Matching results

Data and evaluation

Dataset analysis code/business_entity_resolution/src/eda.py
Validation split code/business_entity_resolution/src/evaluation/make_split.py
Split ID files code/business_entity_resolution/splits
F0.5 scoring code/business_entity_resolution/src/evaluation/f05.py
Training labels
Predictions

Submission checks

Submission validator utils/validate_submission.py
Test records

Other components

Solution author

Connections 22

Solution author → Source records: provides
Source records → Text normalization: normalizes
Source records → Candidate blocking: blocks
Candidate blocking → Candidate pairs: writes candidates
Candidate blocking → Match features: selects pairs
Match features → Match model: scores pairs
Match model → Submission output: produces matches
Submission output → Matching results: writes results
Solution author → Dataset analysis: runs
Dataset analysis → Source records: reads
Solution author → Validation split: runs
Validation split → Source records: reads Source 1
Validation split → Training labels: reads labels
Validation split → Split ID files: writes IDs
Solution author → F0.5 scoring: runs
F0.5 scoring → Training labels: reads
F0.5 scoring → Predictions: reads
Solution author → Submission validator: runs
Submission validator → Test records: reads Source 1
Submission validator → Matching results: checks
Submission validator → Candidate pairs: checks optionally
Test records → Submission validator: checks IDs optionally

Drawn by GitDiagram on September 25, 2026 from the repository's file tree, README and source files. Read as Markdown ·