BQML Models

Train BigQuery ML models inside your warehouse for propensity scoring, LTV prediction, and customer segmentation.

Last reviewed September 15, 2026

BQML Models let you train machine learning models directly inside your BigQuery warehouse with calculated properties from Data Studio. No data leaves your project, and training runs as a standard BigQuery job.

Overview

You select features and a model type. Vendo does the rest:

Generates the CREATE MODEL statement.
Runs it in BigQuery.
Tracks the training status.
Shows evaluation metrics when training completes.

Trained models can feed downstream workflows, for example Value-Based Bidding and segment creation.

Model Types

Vendo supports 14 BigQuery ML model types:

Model Type	Category	Use Cases
Logistic Regression	Classification	Purchase propensity, churn prediction, conversion likelihood
Linear Regression	Continuous prediction	Customer lifetime value (LTV), revenue forecasting
K-Means Clustering	Unsupervised segmentation	Discover natural customer groups, behavioral clustering
Boosted Tree Classifier	Classification	XGBoost classification with higher accuracy
Random Forest Classifier	Classification	Ensemble trees that resist overfitting
Deep Neural Network Classifier	Classification	Neural network for complex patterns
AutoML Classifier	Classification	Selects the best classification model automatically
Boosted Tree Regressor	Continuous prediction	XGBoost regression with higher accuracy
Random Forest Regressor	Continuous prediction	Ensemble trees for robust regression
Deep Neural Network Regressor	Continuous prediction	Neural network for complex numeric prediction
AutoML Regressor	Continuous prediction	Selects the best regression model automatically
ARIMA+	Time series	Time series forecasting with seasonality
Matrix Factorization	Recommendations	Collaborative filtering for recommendations
PCA	Dimensionality reduction	Principal Component Analysis

The Analytics template category includes the Forecasting template, which uses ARIMA+.

Creating a BQML Model

Open Studio.
Click New Data Model.
Select BigQuery ML.
Choose a model type from the Model Types table.
In Features to Include, select the Customer 360 fields. These are pre-computed behavioral metrics, for example order_count, days_since_last_order, and total_revenue.
Configure parameters for the model type that you chose:
- Classification / Regression: select the target label column, and set the train/test split ratio.
- K-Means: choose the number of clusters (2—10).
Click Create Model.
On the model detail page, click Train Model.

Training

Training runs as a BigQuery ML job inside your warehouse. You can monitor progress from the model detail page.

Status	Description
Pending	The model has been created and is queued for training.
Training	BigQuery is actively training the model.
Completed	Training finished successfully. Results and metrics are available.
Failed	Training encountered an error. Check the error message for details.

Training time depends on dataset size and model complexity. Most models on typical e-commerce datasets complete within a few minutes.

Automated Retraining via Pipeline

You can connect BQML models to the orchestration graph as downstream nodes. When calculated property models finish computing, they automatically tell the BQML model to retrain with the latest data:


Source Import  -->  Property Compute  -->  BQML Model Retrain

To set this up, add a pipeline edge from your property compute node to your BQML model node in the Pipelines view.

Results & Evaluation

When training completes, the model detail page shows evaluation metrics and visualizations.

Classification Models (Logistic Regression)

Metric	Description
Accuracy	Percentage of correct predictions
Precision	Proportion of positive predictions that are correct
Recall	Proportion of actual positives that are correctly identified
F1 Score	Harmonic mean of precision and recall
ROC AUC	Area under the receiver operating characteristic curve
Log Loss	Logarithmic loss measuring prediction confidence

Classification results also include:

Feature importance ranking that shows which features contributed most to predictions.
Decile distribution that puts predicted scores into 10 equal groups. It shows how well the model separates high-probability users from low-probability users.

Regression Models (Linear Regression)

Metric	Description
MAE	Mean absolute error
MSE	Mean squared error
RMSE	Root mean squared error
R-squared	Proportion of variance explained by the model

Regression results include feature importance ranking.

Clustering Models (K-Means)

The K-Means section below describes K-Means results.

Value-Based Bidding

You can connect classification models (Logistic Regression) to the Value-Based Bidding (VBB) workflow in Vendo. VBB uses the predicted scores of the model to assign conversion values. Vendo sends these values to ad platforms, so that their bidding algorithms can optimize for your highest-value users.

To connect a model to VBB:

Open the completed classification model.
Open the Value-Based Bidding page of the model.
Link VBB to the destination ad platform.
Set the value mapping for the prediction deciles.

Vendo generates per-user conversion values from the model scores and syncs them on each export cycle.

See Value-Based Bidding for more on the VBB workflow.

K-Means Clustering

K-Means models discover natural groupings in your customer data without a target label. This helps you understand behavioral segments that you may not have defined manually.

Cluster Results

After training, the model detail page shows:

Detail	Description
Cluster count	The number of clusters (as configured)
Cluster sizes	Number of users in each cluster
Top defining features	The features that most distinguish each cluster from others
Davies-Bouldin index	A measure of cluster separation (lower is better)

Visualizations

K-Means results include 2 chart types:

Radar chart that shows the normalized feature values for each cluster, so you can quickly compare clusters side by side.
Trend charts that show how cluster membership and feature distributions change over time when you retrain the model.

Converting Clusters to Segments

You can convert any cluster that a K-Means model discovers into an audience with one click. See Auto-Segmentation for the guided workflow for this process.

Customer 360: calculated properties used by BQML models
Audiences: audiences that you can create from cluster results
Auto-Segmentation: guided K-Means clustering workflow
SQL Models: pre-process data before training