How arkaaiml/heatiq works
HeatIQ is described as an early-warning and thermal-stress system, but the supplied tree and excerpts show an ML-focused weather-to-forecast pipeline rather than health-risk, vulnerability, or GIS implementations. The principal evidenced workflow prepares ERA5 weather features, forms chronological supervised data, trains and evaluates regression models, then builds a deployment artifact for validated single-row inference. The broader README capabilities are not represented by source modules in this tree. Relationships below follow shown imports, calls, and index references; unsampled behavior is not assumed.
Main entry points: build_deployment_artifact.py Deployment builder · artifact.py Artifact contract · predict.py Single-row prediction · evaluate.py Metrics evaluation
Components 13
Weather preparation
- ERA5 cleaning
ml/preprocessing/clean.py - Daily features
ml/preprocessing/features.py - Targets and splits
ml/preprocessing/supervised.py
Training and evaluation
- Regression baselines
ml/models/baseline.py - XGBoost training
ml/models/train_xgboost.py - Metrics evaluation
ml/models/evaluate.py - Deployment builder
ml/models/build_deployment_artifact.py
Inference and artifacts
- Artifact contract
ml/inference/artifact.py - Single-row prediction
ml/inference/predict.py - Model artifact file
Other components
- ML operator
- ERA5 NetCDF
- Temperature forecast
Connections 13
- ERA5 NetCDF → ERA5 cleaning: loads and validates
ml/preprocessing/clean.py - ERA5 cleaning → Daily features: provides weather data
ml/preprocessing/features.py - ML operator → Deployment builder: runs build
ml/models/build_deployment_artifact.py - Deployment builder → Regression baselines: fits regressor
ml/models/build_deployment_artifact.py - Deployment builder → Metrics evaluation: evaluates predictions
ml/models/build_deployment_artifact.py - Deployment builder → Artifact contract: saves artifact
ml/models/build_deployment_artifact.py - Regression baselines → Metrics evaluation: computes metrics
ml/models/baseline.py - XGBoost training → Metrics evaluation: computes metrics
ml/models/train_xgboost.py - Artifact contract → Single-row prediction: delegates prediction
ml/inference/artifact.py - Artifact contract → Model artifact file: reads or writes
ml/inference/artifact.py - Single-row prediction → Temperature forecast: returns prediction
ml/inference/predict.py - ML operator → XGBoost training: runs training
- ML operator → Single-row prediction: requests forecast