How to beginner · 3 min read

How to use MLflow for experiment tracking

Q: How to use MLflow for experiment tracking

Use MLflow to track machine learning experiments by logging parameters, metrics, and artifacts within an mlflow.start_run() context. This enables centralized experiment tracking, comparison, and reproducibility.

Quick answer

Use MLflow to track machine learning experiments by logging parameters, metrics, and artifacts within an mlflow.start_run() context. This enables centralized experiment tracking, comparison, and reproducibility.

PREREQUISITES

Python 3.8+
pip install mlflow>=2.0
Basic knowledge of Python and machine learning

Setup

Install mlflow via pip and set up a local tracking server or use the default local file-based tracking. No environment variables are required for local tracking.

bash

pip install mlflow

Step by step

This example shows how to track an experiment by logging parameters, metrics, and artifacts using mlflow in Python.

python

import mlflow
import mlflow.sklearn
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

# Load data
iris = load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, test_size=0.2, random_state=42)

# Start MLflow run
with mlflow.start_run():
    # Define and train model
    clf = RandomForestClassifier(n_estimators=100, max_depth=3, random_state=42)
    clf.fit(X_train, y_train)

    # Predict and evaluate
    preds = clf.predict(X_test)
    acc = accuracy_score(y_test, preds)

    # Log parameters
    mlflow.log_param("n_estimators", 100)
    mlflow.log_param("max_depth", 3)

    # Log metric
    mlflow.log_metric("accuracy", acc)

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

    print(f"Logged run with accuracy: {acc:.4f}")

output

Logged run with accuracy: 1.0000

Common variations

Use mlflow.start_run(run_name="my_run") to name runs.
Set up a remote tracking server by configuring MLFLOW_TRACKING_URI environment variable.
Log additional artifacts like plots or datasets with mlflow.log_artifact().
Use mlflow.projects to run reproducible projects.

python

import os
import mlflow

# Set remote tracking URI
os.environ["MLFLOW_TRACKING_URI"] = "http://your-mlflow-server:5000"

with mlflow.start_run(run_name="remote_run"):
    mlflow.log_param("example", "remote")
    mlflow.log_metric("metric", 0.9)
    print("Logged to remote tracking server")

output

Logged to remote tracking server

Troubleshooting

If you see ConnectionError, verify your MLFLOW_TRACKING_URI and network access.
If runs do not appear in UI, check that the tracking server is running and accessible.
For permission errors, ensure correct file system permissions or server authentication.

✅

Key Takeaways

Use mlflow.start_run() to create experiment runs and log parameters, metrics, and models.
Set MLFLOW_TRACKING_URI to switch between local and remote tracking servers.
Log artifacts and models to keep all experiment outputs organized and reproducible.

Verified 2026-04

Verify ↗