MLflow Experiment Tracking

Why Track Experiments?

Every data scientist has been there: you run 50 model variations, forget which parameters produced the best result, and end up re-running everything. MLflow solves this.

Setting Up MLflow

pip install mlflow

Start the MLflow UI:

mlflow ui --port 5000

Open http://localhost:5000 to see your experiment dashboard.

Logging Your First Experiment

import mlflow
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_wine
from sklearn.model_selection import train_test_split

# Load data
wine = load_wine()
X_train, X_test, y_train, y_test = train_test_split(
    wine.data, wine.target, test_size=0.2
)

# Start an MLflow run
with mlflow.start_run():
    # Log parameters
    n_estimators = 100
    max_depth = 10
    mlflow.log_param("n_estimators", n_estimators)
    mlflow.log_param("max_depth", max_depth)

    # Train model
    model = RandomForestClassifier(
        n_estimators=n_estimators,
        max_depth=max_depth
    )
    model.fit(X_train, y_train)

    # Log metrics
    accuracy = model.score(X_test, y_test)
    mlflow.log_metric("accuracy", accuracy)

    # Log model
    mlflow.sklearn.log_model(model, "model")

    print(f"Accuracy: {accuracy:.4f}")

Using Autologging

MLflow can automatically log everything:

mlflow.autolog()

with mlflow.start_run():
    model = RandomForestClassifier(n_estimators=100)
    model.fit(X_train, y_train)
    # Parameters, metrics, and model are logged automatically!

Comparing Runs in the UI

The MLflow UI lets you: - Compare metrics across runs side by side - Visualize parameters vs metrics in scatter plots - Sort and filter runs by any metric - Download artifacts from any run

Best Practices

1. Name your experiments — don't use the default experiment
2. Log everything — parameters, metrics, artifacts, tags
3. Use autologging — it captures what you'd forget
4. Tag your runs — add context like "baseline" or "production candidate"
5. Set a tracking URI — use a shared server for team collaboration

From Tracking to Deployment

Once you've found your best model, the next step is deploying it. Learn the complete pipeline — from experiment tracking to Kubernetes deployment — in our MLflow for Kubernetes course.

---

Ready to go deeper? Check out our hands-on course: MLflow for Kubernetes — practical exercises you can follow along on your own machine.