How trangkhanh-ai/steel-industry-energy-consumption works
This repository publishes a data-preparation workflow for forecasting a steel plant’s total electricity use over the next 60 minutes; it does not include model training or deployment. The main workflow validates and chronologically orders the UCI readings, builds leakage-aware features and future targets, assigns time-based splits, then writes processed datasets and train-only exploratory figures. A separate audit script checks feature and leakage conditions, while a sequence utility exposes past-only windows for later deep-learning comparisons. The graph follows documented orchestration and source excerpts; it does not infer connections for unsampled internals.
Main entry points: preprocessing.py Preprocessing · prepare_steel_data.py Preparation runner · Steel_industry_data.csv UCI readings · sequences.py Sequence windows
Components 11
Source validation
- UCI readings
data/steel/raw/Steel_industry_data.csv
Forecast preparation
- Preprocessing
src/steel/preprocessing.py - Preparation runner
scripts/prepare_steel_data.py - Targets and features
src/steel/preprocessing.py - Time-based splits
src/steel/preprocessing.py
Preparation outputs
- Processed datasets
data/steel/processed - Train-only exploration
scripts/prepare_steel_data.py - EDA figures
reports/steel/figures
Audit and sequences
- Feature audit
scripts/audit_features_and_leakage.py - Sequence windows
src/steel/sequences.py
Other components
- Project analyst
Connections 13
- Project analyst → Preparation runner: runs preparation
README.md - Preparation runner → Preprocessing: loads and transforms
scripts/prepare_steel_data.py - UCI readings → Preprocessing: reads readings
- Preprocessing → Targets and features: builds targets
scripts/prepare_steel_data.py - Preprocessing → Time-based splits: splits by time
scripts/prepare_steel_data.py - Preparation runner → Processed datasets: writes datasets
scripts/prepare_steel_data.py - Preparation runner → Train-only exploration: generates exploration
scripts/prepare_steel_data.py - Train-only exploration → EDA figures: writes figures
scripts/prepare_steel_data.py - Feature audit → Processed datasets: checks splits
scripts/audit_features_and_leakage.py - Feature audit → UCI readings: checks source
scripts/audit_features_and_leakage.py - Preprocessing → Sequence windows: provides shared definitions
src/steel/sequences.py - UCI readings → Sequence windows: supplies history
src/steel/sequences.py - Targets and features → Sequence windows: supplies targets
src/steel/sequences.py