How dimpesh10/carbon-emission- works
This repository implements a carbon-emission forecasting project: it synthesizes environmental and activity features, prepares time-series inputs, trains forecasting models, and presents predictions and analysis. The principal user workflow is entering conditions in the Streamlit dashboard and viewing a prediction; the API offers a separate prediction interface for external clients. The graph preserves the documented CNN, Transformer, baseline comparison, evaluation, and SHAP capabilities. Several stage-to-stage handoffs are inferred from shared generated files rather than shown directly, so those links are dashed and unverified; preprocessing and API server internals were not sampled.
Main entry points: 05_cnn_model.py CNN forecaster · 05b_transformer_model.py Transformer forecaster · 04_data_preprocessing.py Time-series preparation · 07_streamlit_app.py Streamlit dashboard
Components 14
Data preparation
- Feature synthesis
carbon_cnn_project/02_data_synthesis.py - Data exploration
carbon_cnn_project/03_eda_visualization.py - Time-series preparation
carbon_cnn_project/04_data_preprocessing.py
Forecast models
- CNN forecaster
carbon_cnn_project/05_cnn_model.py - Transformer forecaster
carbon_cnn_project/05b_transformer_model.py
Prediction interfaces
- Streamlit dashboard
carbon_cnn_project/07_streamlit_app.py - Prediction API
carbon_cnn_project/09_fastapi_server.py - API client simulation
carbon_cnn_project/09_test_api.py
Model analysis
- Baseline comparison
carbon_cnn_project/10_baseline_comparison.py - Model evaluation
carbon_cnn_project/06_model_evaluation.py - SHAP explanations
carbon_cnn_project/08_explainable_ai.py - Report generation
generate_report.py
Other components
- Dashboard user
- API client
Connections 13
- Feature synthesis → Time-series preparation: provides source data
- Time-series preparation → CNN forecaster: prepares training data
- Time-series preparation → Transformer forecaster: prepares training data
- CNN forecaster → Model evaluation: loads checkpoint
carbon_cnn_project/06_model_evaluation.py - Transformer forecaster → Model evaluation: loads checkpoint
carbon_cnn_project/06_model_evaluation.py - CNN forecaster → Baseline comparison: loads checkpoint
carbon_cnn_project/10_baseline_comparison.py - Transformer forecaster → Baseline comparison: loads checkpoint
carbon_cnn_project/10_baseline_comparison.py - Transformer forecaster → SHAP explanations: loads preferred model
carbon_cnn_project/08_explainable_ai.py - CNN forecaster → SHAP explanations: fallback model
carbon_cnn_project/08_explainable_ai.py - Dashboard user → Streamlit dashboard: enters conditions
carbon_cnn_project/07_streamlit_app.py - API client → Prediction API: posts prediction request
- Streamlit dashboard → CNN forecaster: uses fallback model
- Streamlit dashboard → Transformer forecaster: uses primary model
carbon_cnn_project/07_streamlit_app.py