Industrial Clean Energy Offtake Trajectory Classifier using Scikit-Learn
China's systemic carbon emissions reversal is driven by large industrial consumers transitioning power purchase agreements (PPAs) from coal thermal grids to dedicated clean energy projects. Corporate sustainability auditors need automated classifiers to verify whether industrial facility emissions trajectories reflect actual renewable clean energy replacement or cyclical manufacturing downturns. In this challenge, you will build a machine learning classification model using Scikit-Learn. The system will process monthly power consumption features, thermal generation ratios, regional clean energy capacity additions, and economic production indexes. Your solution will classify corporate facilities into structural transitioners vs cyclical operational shifts, providing auditable metrics for green bond issuers and carbon registries.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a Scikit-Learn classification pipeline to categorize whether corporate emissions drops reflect structural clean energy transitions or cyclical downturns.
How work is evaluated
Evaluates classifier accuracy and ROC-AUC on holdout industrial emissions dataset.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
pipeline_export_check
Scikit-Learn pipeline correctly serializes and deserializes via joblib
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
classification_f1_score
Macro F1-score on holdout transition classification task • target: 0.88 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Construct Scikit-Learn Pipelines including preprocessors and classifiers.
Engineer domain features combining energy intensity and economic output indices.
Classify industrial emissions trends into 'Structural Clean Energy Transition' vs 'Cyclical Slump'.
Evaluate precision, recall, and ROC-AUC metrics on holdout test sets.
Reference links and supporting material
Synthetic dataset of industrial facility operational logs including solar PPA fractions, grid electricity draw, and manufacturing output indexes.
Challenge setup
Evaluates classifier accuracy and ROC-AUC on holdout industrial emissions dataset.
Challenge input
JSON with facility_id, solar_ppa_share, thermal_drawdown_pct, production_index
Scikit-Learn
Standard Python library for classical machine learning classification.
Evaluated output
JSON with transition_type, confidence_score, structural_flag (bool)
- Scikit-Learn pipeline correctly serializes and deserializes via joblib
- Macro F1-score on holdout transition classification task • target: 0.88 • range: 0-1
- Benchmark: CleanTransitionBench
- Classification F1 Score target: 0.88
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
- LMDeploy · Required
- RunPod · Optional
- Pieces for Developers · Optional
Evaluation contract
- pipeline_export_check · Weight 1
- classification_f1_score · Weight 1
Metrics
- classification_f1_score · Target 0.88
Benchmarks
- CleanTransitionBench
- energy
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with facility_id, solar_ppa_share, thermal_drawdown_pct, production_index
Output: JSON with transition_type, confidence_score, structural_flag (bool)
Execution harness: Python
Evaluation tools
- Scikit-Learn · Tabular ML Framework
Recipe freeze marker
No recipe freeze time is recorded.