This quickstart walks you through the process of using AutoML Tables web application to do the following steps:
- Create a dataset.
- Import table data from a CSV file into the dataset.
- Identify schema columns in the imported data.
- Train a model from the imported data.
- Use the model to make predictions.
The entire process takes a couple of hours to complete. Most of that time is not active time; you can close your browser window and return to the task later.
Before you begin
Create a project and enable AutoML Tables
Sign in to your Google Account.
If you don't already have one, sign up for a new account.
In the Cloud Console, on the project selector page, select or create a Cloud project.
Make sure that billing is enabled for your Google Cloud project. Learn how to confirm billing is enabled for your project.
- Enable the Cloud AutoML and Storage APIs.
Create a dataset and train a model
Visit AutoML Tables in the Google Cloud Console to begin the process of creating your dataset and training your model.
Select Datasets, and then select New dataset.
Quickstart_Datasetfor the dataset name and click Create dataset.
On the Import your data page, choose Select a CSV file from Cloud Storage.
Leave the Location set to
cloud-ml-tables-data/bank-marketing.csvfor the bucket.
The dataset import takes a few minutes to complete.
After the dataset import completes, select
Depositfor the Target column.
The target column identifies the value the model will be trained to predict.
This window provides information about your imported data. You can click individual rows to see more about distribution and correlation for a specific feature.
Click Train model. Enter
Quickstart_Modelfor Model name, and
1for Training budget.
Click Train model to start the training process.
Model training takes about two hours to complete. After the model is successfully trained, the Models tab shows high-level metrics for the model.
Select the Evaluate tab for a detailed view of the model evaluation metrics.
For this model,
1represents a negative outcome--a deposit is not made at the bank.
2represents a positive outcome--a deposit is made at the bank.
You can select a label to see specific evaluation metrics for that label. You can also adjust the Score threshold to see how the metrics differ for different threshold values.
You can also scroll down to see the confusion matrix and feature importance graph.
Select the Test & Use tab, and select Online prediction.
Click Deploy model to deploy your model.
You must deploy your model before you can request online predictions. Deploying a model takes a few minutes to complete.
When the model is deployed, AutoML Tables fills in sample data to help you test your model.
Select the Generate feature importance checkbox.
Click Predict to request the online prediction.
AutoML Tables determines the probability of each possible outcome based on the input values and displays the confidence values for the prediction in the Prediction result section.
In the example above, the model is predicting the result of "1", with 99.8% certainty.
You can also submit prediction requests in batch form. Learn more.
If you no longer need your custom model or dataset, you can delete them.
To avoid unnecessary Google Cloud Platform charges, use the Cloud Console to delete your project if you do not need it.
Undeploy your model
Your model incurs charges while it is deployed.
- Select Models and click on the model you want to undeploy.
- Select the Test & Use tab and click Online prediction.
- Click Remove deployment.
Delete a model
To delete a model, select Models. Click the More actions menu for the model that you want to delete, and then select Delete model.
Delete a dataset
To delete a dataset, select Datasets. Click the More actions menu for the model that you want to delete, and then select Delete dataset.
- Learn about AutoML Tables features and capabilities.
- Run through more examples using Colab notebooks.
- Read through our Beginner's guide.
- Get started preparing your training data for your own AutoML Tables deployment.
- Learn about interpreting prediction results.
- Learn about local feature importance.