Machine learning opens up opportunities to get more value out of data, and business users are eager to see that value. However, today’s machine learning experts are facing a lot of requests and their expertise is at a premium. What if data analysts had the ability to create and test their own machine learning models?
At Google Cloud, we think this would allow analysts to accelerate the adoption of predictive analytics into their business intelligence. Using governed data in your Looker environment to build ML models can help transform the insights you gain across all aspects of your business – and make those insights available to your entire team.
From our partners:
Earlier this year we released the Machine Learning (ML) Accelerator for Looker, a Looker application that integrates Looker with BigQuery ML. It streamlines machine learning workflows by providing simplified access to machine learning tools and letting users use trusted data to train and run those models. The application handles all steps of model creation, from training to evaluation to inference generation, all based on the “source of truth” data in Looker. The opportunity to use ML without having to write any code and in a Looker user’s current business intelligence platform will encourage deeper adoption of machine learning by data analysts. The Machine Learning Accelerator is available free from the Looker Marketplace, and can be installed in any Looker instance.
How it works
The ML Accelerator for Looker is fully integrated with the Google Cloud environment, allowing business users that rely on Looker, analyst teams who work with BigQuery, and data scientists who rely on Vertex AI to work together. For example, a business user recognizes a need, and evaluates several models until they find one that meets their needs. They can then ask the BigQuery data team for additional data to enhance the model. Then finally, a data scientist can carry it forward into production to start using the model at scale.
Today, we are adding a new ML model type: Single Time Series Forecasting using the ARIMA Plus model in BigQuery ML. This is one of the most popular model types in BigQuery ML, and we are excited to bring it to Looker users. Simply select a Looker Explore, choose any metric and its corresponding time dimension, and soon you’ll have a forecast for that metric. You can use the Time Series Forecasting model type to predict future trends for business metrics such as product sales or total revenue.
ML Accelerator for Looker also supports Classification and Regression model types, which cover a wide variety of use cases. Analysts researching Looker data about past customer churn can take those same Looker data sets to predict future churn risks using a classification model. Or, a marketing team with data on the return on advertising spend (ROAS) of past marketing campaigns can use a regression model to predict the ROAS of future campaigns. Both of those model types use the Boosted Tree technique, which comes with Explainable AI, so they can even uncover which characteristics are most important to predict the outcome. Our roadmap includes support for more BigQuery ML model types, but let us know what you’re interested in by contacting your Looker account team!
By: Andy Crutchfield (Cloud Consultant) and Chris Schmidt (Product Technology Manager)
Originally published at: Google Cloud Blog
For enquiries, product placements, sponsorships, and collaborations, connect with us at firstname.lastname@example.org. We'd love to hear from you!
Our humans need coffee too! Your support is highly appreciated, thank you!