How sweezalsuares/nexusretail-ai-powered-e-commerce-customer-intelligence works
NexusRetail trains three order-level classifiers—profitability, delivery speed, and return risk—and exposes their results through a Streamlit dashboard and prediction interface. Training loads retail orders, creates targets and engineered features, selects features, fits ANN and gradient-boosting pipelines, evaluates them, and saves model and report artifacts. Users explore dataset and model diagnostics, adjust thresholds, or submit single and batch orders for predictions. The dashboard and prediction workflow are documented; some dashboard internals are only partially excerpted, so edges are limited to documented or indexed references.
Main entry points: train.py Training orchestrator · data_prep.py Targets and features · predict.py Order prediction · streamlit_app.py Retail dashboard
Components 14
Model training
- Project settings
config.py - Targets and features
src/data_prep.py - Feature selection
src/feature_selection.py - ANN and HGB pipelines
src/models.py - Training orchestrator
src/train.py - Model evaluation
src/evaluate.py - Static diagnostics
src/plots.py
Prediction
- Order prediction
src/predict.py
Dashboard
- Retail dashboard
app/streamlit_app.py
Data and artifacts
- Order dataset
data/ecommerce_sales_34500.csv - Saved pipelines
models - Metrics and predictions
reports
Other components
- Retail user
- CLI user
Connections 21
- Retail user → Retail dashboard: explores and submits
README.md - CLI user → Order prediction: scores orders
README.md - Targets and features → Order dataset: loads orders
src/data_prep.py - Targets and features → Project settings: uses data settings
src/data_prep.py - Training orchestrator → Project settings: uses training settings
src/train.py - Training orchestrator → Targets and features: loads and transforms
src/train.py - Training orchestrator → Feature selection: selects features
src/train.py - Feature selection → Targets and features: uses feature definitions
src/feature_selection.py - Training orchestrator → ANN and HGB pipelines: builds classifiers
src/train.py - ANN and HGB pipelines → Targets and features: engineers features
src/models.py - Training orchestrator → Model evaluation: computes metrics
src/train.py - Training orchestrator → Static diagnostics: creates figures
src/train.py - Training orchestrator → Saved pipelines: saves pipelines
src/train.py - Training orchestrator → Metrics and predictions: writes metrics and predictions
src/train.py - Static diagnostics → Metrics and predictions: saves figures
src/plots.py - Order prediction → Saved pipelines: loads pipelines
src/predict.py - Order prediction → Metrics and predictions: reads thresholds
src/predict.py - Order prediction → Targets and features: validates input schema
src/predict.py - Retail dashboard → Order prediction: requests predictions
app/streamlit_app.py - Retail dashboard → Targets and features: loads dataset
app/streamlit_app.py - Retail dashboard → Project settings: uses app settings
app/streamlit_app.py