Vertex AI AutoML: Tutorial to create classification models
Build high quality models with your datasets without writing any code.
In today’s data-driven world, organizations are constantly seeking ways to harness the power of their data to make informed decisions and gain a competitive edge. Machine learning has emerged as a crucial tool in this endeavor, allowing businesses to extract valuable insights and predictions from their data. However, machine learning model development can be a process often requiring a high level of AI expertise.
Google Cloud’s AI portfolio has been designed to cater to all kind of users — whether business user, developer or an AI practitioner.
If you are an organisation or user looking to get value out of your rich dataset without having to write any code, you can explore Vertex AI AutoML. AutoML offers a GUI tool with an intuitive guide to help you with the end-to-end process of ML model development and deployment.
The whole process is automated: automatically train your model, automated feature engineering, model selection, tuning before putting it into production and finally an automatic deployment of it with a click to start making predictions.
What used to take months can now be done in weeks or even a matter of days.
📝 Steps to Create Classification Models
Let’s delve into the process of creating classification models using Vertex AI AutoML, highlighting how it simplifies and expedites the entire journey.
🗂️ Creating your dataset
To begin, you need a dataset. For this tutorial, we’ll assume your data is stored in Google Cloud Storage, but you can also upload it directly from your computer or use data stores in BigQuery. Preparing your data is a critical step, and Vertex AI AutoML allows you to work with structured data, images, or text, depending on your project’s requirements.
🛠️ Training your model
Below are the steps to follow once your dataset is created:
- Go to the Google Cloud Console and navigate to Vertex AI > Model Registry.
- Click on the “Create Model” button.
- Select the appropriate dataset you prepared in the previous step.
- Choose to create a model using AutoML for quick and hassle-free model building. Alternatively, you can opt for custom training if you have specific frameworks like TensorFlow or XGBoost in mind. This blog focuses on AutoML so we will use the first option.
- Customize your model easily by specifying which fields to include or exclude from your training data.
- Set the number of nodes per hour for training, starting with a small number to save costs initially. The training time may vary, typically taking around 5–6 hours or more, depending on your dataset’s size and node configuration.
- Once trained, view your model evaluation fields after navigating to your model in the Model Registry to assess its performance.
Deployment and Responsible AI
But Vertex AI AutoML doesn’t stop at model creation. Once your model is ready, you can seamlessly deploy and test it within the same user interface using Vertex AI endpoints. This streamlined deployment process saves you valuable time, ensuring that your models are ready to provide predictions and insights to your organization swiftly.
Understanding Model Outputs with Feature Attribution
One of the key features that Vertex AI AutoML offers is feature attribution. This feature helps organizations adhere to the principles of responsible AI by shedding light on how the model makes decisions and which features contribute most to the model’s outputs. It doesn’t keep the model as a black box, even for someone who is a beginner in this space.
Feature attribution allows you to:
- Interpret Model Outputs: By understanding the contribution of each feature, you gain insights into why the model makes specific predictions. This transparency is essential for responsible AI, as it helps you ensure that your model’s decisions are fair and unbiased.
- Identify Model Biases: Feature attribution can reveal any biases present in your model by highlighting which features have the most significant impact on predictions. This insight empowers you to take corrective measures and align your model with ethical standards.
Use Cases Beyond Classification
While we used Vertex AI AutoML for classification in this blog, it’s essential to note that it’s a versatile tool that can be applied to various machine learning tasks. Here are some additional use cases:
- Forecasting: You can use Vertex AI AutoML for time series forecasting, helping you predict future trends and make informed decisions based on historical data.
- Regression Models: It’s equally adept at building regression models, which are used to predict continuous numerical values, making it valuable for tasks like price prediction and demand forecasting.
In conclusion, Google Cloud’s Vertex AI AutoML simplifies the process of building machine learning models while promoting responsible AI practices. Its ease of use, automation, and feature attribution reduce the barriers to entry for those without extensive machine learning expertise. To learn more about Vertex AI, head to the these tutorials and play with image, text, tabular or video format datasets.
Happy building! 🖥️
