// MLOps Guides 2026 ⏱ ~3 min read

MLflow Tutorial: How to Track Machine Learning Experiments (2026)

Ayub Shah

· 📅 April 2026 · 👤 ML engineers & data scientists

⚡ Quick Answer

This MLflow tutorial teaches you to track machine learning experiments in 20 minutes. Learn to log parameters, metrics, and models; compare runs side-by-side; and reproduce any result. Includes copy-paste code for scikit-learn.

87%of ML projects never reach production

20minto complete this tutorial

100%free & open source

Looking for a hands-on MLflow tutorial that actually shows you how to track machine learning experiments? This step-by-step guide will have you logging parameters, metrics, and models in under 20 minutes — with real code you can copy and run today.

This MLflow tutorial is designed for ML engineers and data scientists who want to stop losing experiment results and start reproducing winning models. No prior MLflow experience required.

Table of Contents

01What is MLflow?Essential
02PrerequisitesSetup
03Install MLflowStep 1
04Start Tracking ServerStep 2
05Write Training ScriptStep 3
06View Results in UIStep 4
07Compare Multiple RunsStep 5
08What’s Next?Advanced

01 What is MLflow?

MLflow is an open-source platform that tracks everything about your machine learning experiments — parameters, metrics, model artifacts, and code versions — so you can reproduce any result and never lose a winning configuration again.

In this MLflow tutorial, you'll learn the core concepts that make MLflow the industry standard for experiment tracking, used by thousands of companies including Databricks, Roku, and Zillow.

💡

WHY THIS MATTERS

Without experiment tracking, most ML engineers waste hours rerunning experiments they've already done — or ship models they can't reproduce. MLflow eliminates both problems permanently.

02 Prerequisites

🐍

Python 3.8+

Run python --version to check

📦

pip installed

Comes with Python 3.4+

🧠

Basic ML knowledge

Know what training and accuracy mean

This MLflow tutorial assumes you're comfortable with basic Python and scikit-learn. No prior MLOps experience needed.

03 Install MLflow

⏱ 2 minutes

MLflow is a single pip install. It includes the tracking server, UI, and Python API — no Docker, no cloud setup required.

BASHterminal
$ pip install mlflow scikit-learn

Collecting mlflow
  Downloading mlflow-2.x.x-py3-none-any.whl (24.6 MB)
...
Successfully installed mlflow-2.x.x scikit-learn-1.x.x

$ mlflow --version
mlflow, version 2.x.x

				
					
				1
2
3
4
5
6
7
8
9

						$ pip install mlflow scikit-learn
 
Collecting mlflow
  Downloading mlflow-2.x.x-py3-none-any.whl (24.6 MB)
...
Successfully installed mlflow-2.x.x scikit-learn-1.x.x
 
$ mlflow --version
mlflow, version 2.x.x

					

			

✅

USING A VIRTUAL ENVIRONMENT?

Run python -m venv .venv && source .venv/bin/activate before installing.

04 Start the MLflow Tracking Server

⏱ 1 minute

BASHterminal
$ mlflow ui

[2026-04-15 10:23:01 +0000] [12345] [INFO] Starting gunicorn 21.2.0
[2026-04-15 10:23:01 +0000] [12345] [INFO] Listening at: http://127.0.0.1:5000

				1
2
3
4

						$ mlflow ui
 
[2026-04-15 10:23:01 +0000] [12345] [INFO] Starting gunicorn 21.2.0
[2026-04-15 10:23:01 +0000] [12345] [INFO] Listening at: http://127.0.0.1:5000

⚠️

PORT CONFLICT?

If port 5000 is taken, run mlflow ui --port 5001 instead.

05 Write Your First MLflow Tracking Script

⏱ 10 minutes

Create a file called train.py. This is the core of our MLflow tutorial — you'll see exactly how to log parameters, metrics, and models.

PYTHONtrain.py
import mlflow
import mlflow.sklearn
from sklearn.datasets import load_iris
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, f1_score

# Configuration
N_ESTIMATORS = 100
MAX_DEPTH = 5
RANDOM_STATE = 42

# 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=RANDOM_STATE
)

# MLflow: name your experiment
mlflow.set_experiment("iris-classifier")

with mlflow.start_run():
    # Train model
    model = RandomForestClassifier(
        n_estimators=N_ESTIMATORS,
        max_depth=MAX_DEPTH,
        random_state=RANDOM_STATE
    )
    model.fit(X_train, y_train)

    # Generate predictions
    predictions = model.predict(X_test)
    accuracy = accuracy_score(y_test, predictions)
    f1 = f1_score(y_test, predictions, average="weighted")

    # Log parameters, metrics, and model
    mlflow.log_param("n_estimators", N_ESTIMATORS)
    mlflow.log_param("max_depth", MAX_DEPTH)
    mlflow.log_metric("accuracy", accuracy)
    mlflow.log_metric("f1_score", f1)
    mlflow.sklearn.log_model(model, "random-forest-model")

    print(f"Accuracy: {accuracy:.4f} | F1: {f1:.4f}")
    print(f"Run ID: {mlflow.active_run().info.run_id}")

				
					
				1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

						import mlflow
import mlflow.sklearn
from sklearn.datasets import load_iris
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, f1_score
 
# Configuration
N_ESTIMATORS = 100
MAX_DEPTH = 5
RANDOM_STATE = 42
 
# 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=RANDOM_STATE
)
 
# MLflow: name your experiment
mlflow.set_experiment("iris-classifier")
 
with mlflow.start_run():
    # Train model
    model = RandomForestClassifier(
        n_estimators=N_ESTIMATORS,
        max_depth=MAX_DEPTH,
        random_state=RANDOM_STATE
    )
    model.fit(X_train, y_train)
 
    # Generate predictions
    predictions = model.predict(X_test)
    accuracy = accuracy_score(y_test, predictions)
    f1 = f1_score(y_test, predictions, average="weighted")
 
    # Log parameters, metrics, and model
    mlflow.log_param("n_estimators", N_ESTIMATORS)
    mlflow.log_param("max_depth", MAX_DEPTH)
    mlflow.log_metric("accuracy", accuracy)
    mlflow.log_metric("f1_score", f1)
    mlflow.sklearn.log_model(model, "random-forest-model")
 
    print(f"Accuracy: {accuracy:.4f} | F1: {f1:.4f}")
    print(f"Run ID: {mlflow.active_run().info.run_id}")

					

			

Run the script: python train.py

✅

IT WORKED!

MLflow created an mlruns/ folder with everything it logged — parameters, metrics, and your trained model.

06 View Your Results in the MLflow UI

⏱ 2 minutes

Open your browser and go to http://localhost:5000. You'll see the MLflow UI with your experiment listed.

MLflow UI showing metric tracking dashboard

Figure 1: MLflow tracking UI — metrics like accuracy are visualized automatically

07 Compare Multiple Runs Side-by-Side

⏱ 5 minutes

Run the script 3-4 times with different parameters (change n_estimators or max_depth). Then in the MLflow UI, select multiple runs and click "Compare".

MLflow run comparison table

Figure 2: Compare runs side-by-side — MLflow shows which parameters produced the best results

🏆

YOU CAN NOW ANSWER

"Which configuration gave us the best result, and can we reproduce it?" — with a single click. That's the power of this MLflow tutorial in practice.

08 What to Learn Next

📋

Model Registry

Promote your best run to "Production" with one click.

📊

Log More Metrics

Add precision, recall, and per-epoch training loss.

🚀

Serve Your Model

Run mlflow models serve to expose as REST API.

☁️

Remote Tracking Server

Point your team at one shared MLflow server.

📖 External resources: Official MLflow Documentation • MLflow GitHub

📚