sadhana-panthi/freight-rate_ml

How sadhana-panthi/freight-rate_ml works

This repository is a freight-rate prediction workflow. Its primary path prepares training and scoring data, evaluates a time-based holdout, fits a gradient-boosted regression pipeline, and writes validation and December predictions. Separate analysis scripts compare models and features and report data-quality findings; the provided scorer validates deliverables and creates the December chart. The model’s preprocessing and target transformation are implemented inside the main prediction module. Edges reflect the shown calls and documented file handoffs; the CSV contents and unsampled runtime behavior are not inferred.

Main entry points: predict_rates.py Feature preparation · compare_models.py Model comparison · score.py Submission scorer · data_checks.py Data findings

Components 18

Input data

Training loads data/train_test.csv
Validation loads data/validation.csv
December inputs data/december_chart_inputs.csv
ID template data/validation_predictions_template.csv

Prediction workflow

Feature preparation predict_rates.py
Holdout evaluation predict_rates.py
Regression pipeline predict_rates.py
Impute and encode predict_rates.py
Gradient boosting predict_rates.py
Prediction writer predict_rates.py
Validation predictions validation_predictions.csv
December predictions december_predictions.csv

Model analysis

Model comparison compare_models.py
Feature ablation compare_models.py
Data findings data_checks.py

Output checks

Submission scorer score.py
December chart scorer_results

Other components

Analyst

Connections 24

Analyst → Prediction writer: runs README.md
Training loads → Feature preparation: reads predict_rates.py
Validation loads → Feature preparation: reads predict_rates.py
December inputs → Feature preparation: reads predict_rates.py
Feature preparation → Holdout evaluation: prepares features predict_rates.py
Holdout evaluation → Regression pipeline: fits and evaluates predict_rates.py
Prediction writer → Feature preparation: prepares features predict_rates.py
Prediction writer → Regression pipeline: fits model predict_rates.py
Regression pipeline → Impute and encode: applies predict_rates.py
Regression pipeline → Gradient boosting: trains predict_rates.py
ID template → Prediction writer: supplies IDs predict_rates.py
Prediction writer → Validation predictions: writes predict_rates.py
Prediction writer → December predictions: writes predict_rates.py
Analyst → Model comparison: runs README.md
Model comparison → Feature preparation: uses compare_models.py
Model comparison → Regression pipeline: builds models compare_models.py
Model comparison → Feature ablation: computes variants compare_models.py
Analyst → Data findings: runs README.md
Data findings → Feature preparation: uses data_checks.py
Data findings → Regression pipeline: evaluates errors data_checks.py
Analyst → Submission scorer: runs README.md
Validation predictions → Submission scorer: validates README.md
December predictions → Submission scorer: validates README.md
Submission scorer → December chart: creates score.py

More Python diagrams: tomgoldstein/loss-landscape · coreruleset/coreruleset · jy0205/pyramid-flow · roboflow/trackers · sourcebot-dev/sourcebot

Drawn by GitDiagram on October 1, 2026 from the repository's file tree, README and source files. Read as Markdown ·