user@devops:~$ cat README.md
MLflow MLOps
# Description
Complete MLOps project with MLflow on the Wine (UCI) dataset — 178 samples, 13 chemical features, 3 wine classes. Implements MLflow Tracking to log parameters, metrics, and artifacts for 3 models (Logistic Regression, Random Forest, Gradient Boosting) with 5-fold cross-validation and learning curves. MLflow Model Registry versions each model and promotes the best (LogisticRegression, 97.72% CV accuracy) through Staging → Production. Includes inference by loading the model directly from Production, 10 visualizations (EDA, confusion matrices, feature importance, comparisons, learning curves), and a complete reproducible experimentation workflow with SQLite backend.
# Key features
$ MLflow Tracking: parameters, metrics, artifacts per training run
$ 5-fold cross-validation with comparative boxplot between models
$ Feature importance analysis for Random Forest and Gradient Boosting
$ Learning curves to diagnose bias/variance for all 3 models
$ MLflow Model Registry: versioning and stages (None → Staging → Production)
$ Automatic promotion of best model to Production
$ Production inference: load registered model and predict on 5 test samples
$ 10 visualizations: EDA, confusion matrices, comparisons, learning curves
$ SQLite backend for persistent tracking
# Gallery
# Technologies used
