How vedthacker/ecommerce-rto-predictor works
This is a pre-dispatch e-commerce Return-to-Origin (RTO) prediction engine. Its principal workflow is order entry in the Streamlit interface, inference and risk/action presentation; a separate batch workflow generates synthetic orders, explores and engineers data, trains and evaluates a model, scores orders, and analyzes those scores in SQLite. SQL business intelligence and real-time contextual signals are distinct capabilities. The real-time module is present, but the excerpts do not show it being wired into inference. Edges reflect shown calls or README-documented behavior; generated data flow details are limited where artifacts are not in the tree.
Main entry points: main.py Pipeline Runner · model.py Model Training · predictor.py Risk Inference · sql_analysis.py SQL Business Analysis
Components 16
Order Risk Prediction
- Order Entry UI
app.py - Risk Inference
src/predictor.py - Saved Model
outputs/models/rto_model.pkl - Risk and Action
src/predictor.py
Data and Model
- Pipeline Runner
main.py - Synthetic Order Generator
data/generate_data.py - Raw Order Data
data/generate_data.py - Exploratory Analysis
src/eda.py - Feature Engineering
src/preprocess.py - Model Training
src/model.py
Business Analytics
- Scored Orders
src/model.py - SQL Business Analysis
src/sql_analysis.py - SQLite Orders Store
src/sql_analysis.py
Context Signals
- Real-Time Signals
src/realtime_features.py
Other components
- Operations User
- Open-Meteo
Connections 16
- Operations User → Order Entry UI: enters order
app.py - Order Entry UI → Risk Inference: requests prediction
app.py - Risk Inference → Saved Model: loads model
README.md - Risk Inference → Risk and Action: returns score
src/predictor.py - Pipeline Runner → Synthetic Order Generator: runs generation
main.py - Synthetic Order Generator → Raw Order Data: writes orders
data/generate_data.py - Pipeline Runner → Exploratory Analysis: runs analysis
main.py - Pipeline Runner → Model Training: runs training
main.py - Pipeline Runner → SQL Business Analysis: runs SQL analysis
main.py - Exploratory Analysis → Feature Engineering: uses engineering
src/eda.py - Model Training → Feature Engineering: engineers features
src/model.py - Model Training → Saved Model: saves model
src/model.py - Model Training → Scored Orders: writes scores
src/model.py - SQL Business Analysis → Scored Orders: loads scores
src/sql_analysis.py - SQL Business Analysis → SQLite Orders Store: loads and queries
src/sql_analysis.py - Real-Time Signals → Open-Meteo: requests weather
src/realtime_features.py