You’re Just One Article Away from Building Your First ML Model (No Coding Required)

Author(s): Alex Vysotskiy

Originally published on Towards AI.

Zero to ML Hero: A Beginner’s Guide to Vertex AI for Non-Coders

When you hear “Machine Learning,” do you immediately think of complicated algorithms, endless coding, and intimidating data science jargon? Me too — at least, that’s what I thought before trying out Vertex AI by Google.

As someone without a data science degree, I was amazed at how straightforward and accessible building a machine learning (ML) model could be. If you’ve ever dreamed of harnessing ML but were put off by the technical complexity, this article is your ticket to finally jumping in.

Let me show you how simple it can be — no coding required!

🎯 Step 0: Prepare Your Data

Before we dive into Vertex AI, you’ll need some data to work with. We won’t be covering data preparation in this article, but even a simple CSV file can do the trick. For example, your CSV could have columns like:

• Customer_ID (identifier)
• Age (numeric)
• Income (numeric)
• Membership_Status (categorical, e.g., “Silver,” “Gold,” “Platinum”)

Using such data, you could predict a customer’s membership status based on their age and income.

This example is intentionally simplistic for ease of understanding, but rest assured, the exact steps we’ll cover can be applied to much more complex ML problems, such as:

• Predicting customer churn based on user behavior data.
• Estimating house prices based on real estate features.
• Classifying images or text documents into specific categories.

If you don’t have your own dataset ready, you can download this sample CSV here. (Below, you’ll see a preview of what the dummy data looks like.)

🚀 Step 1: Accessing Vertex AI

First, head over to Vertex AI. You’ll need a Google Cloud account, which you can set up in just a few minutes. Don’t worry; Google provides free credits and a 90-day trial, perfect for beginners experimenting with ML.

📂 Step 2: Create a New Dataset

Once you’re inside Vertex AI:

Click on “Datasets” from the menu.

Choose “Create dataset” and pick your data type (e.g., tabular data, images, or text). For the dummy data, choose tabular → regression or classification. This is the most common type of problem when you have certain characteristics of each data point called ‘features’. In our case it is the age and income of a user and we want to predict which membership_status the user will have.

Upload your dataset directly from your device or Google Cloud Storage (see the screenshots below)

That’s it — you’ve successfully created your first dataset in Vertex AI!

🤖 Step 3: Train Your Model

Now comes the fun part — training your model without writing a single line of code:

After uploading your dataset, select “Train new model.”

Just use the default suggestions. But as you can see there is room for all the different customizations and alternative model types. The only thing we need to select in the first three steps is the target column on the “Training method” page. Select ‘membership_status’ or whatever the column name of the result that you are trying to predict.

On the ‘training options’ page it will ask you about the columns that you want to use for training. Remove columns that don’t help with prediction — for example, ‘Customer_ID’ won’t influence outcomes.

Finally, choose 1 hour as your training budget — this is the minimum allowed and more than enough for us.

Vertex AI automatically suggests the most suitable algorithm. Just trust it and proceed!

Grab a coffee; your model is now training.

🔍 Step 4: Evaluate the Results

You can monitor the results in the ‘Training’ section.

When training finishes, you can see the results in the model registry.

Vertex AI provides several metrics to help you evaluate your results. Here are the key ones to know:

• ROC AUC: This score shows how well your model ranks the correct answers higher than the wrong ones, across all possible thresholds. You can think of it like the model’s ability to tell apart different classes — 1.0 is perfect, 0.5 is just guessing. While it’s a useful benchmark, real-world decisions usually depend more on precision and recall.
• Precision: Indicates how many of the predicted positive cases are actually positive (e.g., of customers predicted to be “Gold” members, how many really are).
• Recall: Reflects how many actual positive cases the model correctly identified (e.g., out of all real “Gold” members, how many the model correctly predicted).

You can optimize for precision or recall depending on your business goal. For example, if you’re building a model for airport security to identify whether a bag contains explosives, you’d want to prioritize recall over precision. In other words, it’s better to check more bags from innocent passengers than to risk overlooking even one that contains a threat.

We won’t dive too deep into the technical interpretation of these metrics in this article but as you would expect — higher values typically mean better model performance. Vertex AI also provides clear visualizations to help you quickly grasp how effectively your model is working and compare it to a previous version of the model.

🎉 Step 5: Test and Deploy Your Model

Once satisfied with the results, testing and deploying your model is as easy as a few clicks.

You can predict values for a single user or multiple users using the ‘Batch predict’ tab (see the screenshot above). Manual inputs or CSVs still work here. Just upload the rows similar to what your training data was — but this time you’ll leave out the column with outcomes (like in our case, we would have no column ‘membership_status’). The model will predict the values for you, based on the features.

It is also possible to deploy your model to an endpoint for devs to use it. Check out the ‘Deploy & Test’ tab, also visible on the last screenshot. This falls more into the ‘advanced’ category of questions and we will not cover it here.

Congratulations — you now have your own ML model ready to use!

Wrapping Up

Of course, this is not the only way to build ML models but Vertex AI stands out because:

• No coding required: Perfect for non-coders.
• User-friendly interface: Easy navigation and clear instructions.
• Powerful capabilities: Google’s infrastructure at your fingertips.

Building your first machine learning model doesn’t have to be daunting. Vertex AI made it a fun and enjoyable experience for me personally. And just to be clear — I have no affiliation with Google.

Give it a try — you’re now just moments away from becoming an ML Hero!

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI