How ish-code1/cse-mlops works
This repository trains and serves an Iris classifier. The training workflow loads and splits the Iris dataset, fits and evaluates a random forest, applies an accuracy gate, saves a local joblib artifact, and logs metrics and a model to MLflow. At runtime, FastAPI loads the MLflow-format model during startup; the prediction endpoint validates four positive measurements, invokes the model, and returns a class label. The README is empty, so the map reflects the sampled source; unsampled files may contain additional details.
Main entry points: train.py Training workflow · main.py FastAPI app · MLmodel MLflow model · Iris_random_forest_model.joblib Joblib model
Components 13
Model training
- Training workflow
train.py - Iris dataset
train.py - Random forest
train.py - Evaluation gate
train.py
Model artifacts
- Joblib model
models/Iris_random_forest_model.joblib - MLflow tracking
train.py - MLflow model
model/MLmodel
Prediction API
- FastAPI app
main.py - Model loader
main.py - Input validation
main.py - Prediction endpoint
main.py - Health routes
main.py
Other components
- MLflow service
Connections 13
- Training workflow → Iris dataset: loads and splits
train.py - Training workflow → Random forest: fits classifier
train.py - Training workflow → Evaluation gate: evaluates and gates
train.py - Training workflow → Joblib model: saves model
train.py - Training workflow → MLflow tracking: logs run and model
train.py - MLflow tracking → MLflow service: tracks and registers
train.py - FastAPI app → Model loader: loads at startup
main.py - Model loader → MLflow model: loads model
main.py - Model loader → MLflow service: sets tracking URI
main.py - FastAPI app → Input validation: validates request
main.py - FastAPI app → Prediction endpoint: routes prediction
main.py - Prediction endpoint → MLflow model: invokes predict
main.py - FastAPI app → Health routes: serves health
main.py