How kavurubuvanesh/flyrank-ml-workspace works
FlyRank ML Internship is an educational starter repo for applied search discoverability, with a runnable refresh-opportunity modeling example. The principal workflow is to prepare the anonymized page data, compare a transparent baseline with trained classifiers, evaluate and rank review opportunities, then produce reports for human review. The README also documents Colab notebooks and a gated full-data notebook path. Script relationships are supported by the sampled code and README; notebook internals were not sampled, so their data wiring is not inferred.
Main entry points: 04_evaluate_and_export.py Queue evaluation · run_all.py Pipeline runner · 01_prepare_features.py Prepare features · 02_baseline_score.py Baseline scorer
Components 17
Learning entry points
- First-win notebooks
notebooks - Full-release notebook
notebooks/03_working_with_the_full_release.ipynb
Modeling pipeline
- Pipeline runner
scripts/run_all.py - Prepare features
scripts/01_prepare_features.py - Baseline scorer
scripts/02_baseline_score.py - Model training
scripts/03_train_model.py - Queue evaluation
scripts/04_evaluate_and_export.py - ML utilities
scripts/ml_utils.py
Review outputs
- PDF report builder
scripts/05_build_pdf_report.py - Ranked review queue
outputs - Charts and reports
outputs
Data sources
- Anonymized sample
data/raw/content_refresh_anonymized.csv - Feature vector
- Baseline queue
- Model predictions
- Gated dataset
Other components
- Intern
Connections 25
- Intern → Pipeline runner: runs pipeline
- Pipeline runner → Prepare features: dispatches
- Pipeline runner → Baseline scorer: dispatches
- Pipeline runner → Model training: dispatches
- Pipeline runner → Queue evaluation: dispatches
- Pipeline runner → PDF report builder: dispatches
- Prepare features → Anonymized sample: reads
- Prepare features → Feature vector: writes
- Baseline scorer → Feature vector: reads
- Baseline scorer → Baseline queue: writes
- Model training → Feature vector: reads
- Model training → Baseline queue: reads
- Model training → Model predictions: writes
- Queue evaluation → Feature vector: reads
- Queue evaluation → Baseline queue: reads
- Queue evaluation → Model predictions: reads
- Queue evaluation → Ranked review queue: writes
- Queue evaluation → Charts and reports: writes
- PDF report builder → Charts and reports: writes
- Prepare features → ML utilities: uses helpers
- Baseline scorer → ML utilities: uses helpers
- Queue evaluation → ML utilities: uses helpers
- Intern → First-win notebooks: opens
- Intern → Full-release notebook: opens
- Full-release notebook → Gated dataset: queries full release