Unveiled at AWS re:Invent 2021, Canvas allows workers lacking ML skills to build models that generate actionable insights from their data sets without writing a single line of code. Utilizing a visual interface, Canvas users can browse and access disparate data sources in the cloud, including Amazon S3, Redshift, or on-premises, combine the datasets, train models, and generate predictions.

Canvas utilizes SageMaker to clean, correct, and combine data automatically, then creates models and selects the most accurate by assigning the results an accuracy score. It supports multiple problem types such as binary, multi-class, numerical regression, and time series forecasting. During a presentation at AWS re:Invent 2021 highlighting the capabilities of Canvas, Kimberly Madia, AWS Sr. Manager of Product Marketing, demonstrated how Canvas can help generate confidence in the results. Using a feature known as Analyze View, Canvas can explain the prediction to those using the model, providing answers to questions while allowing additional model review for potential errors.

One of the most challenging chores in any business is to develop accurate predictions or forecasts regarding sales and revenue. Companies spend countless hours gathering data, building models, and then discussing and revising models typically built using Microsoft Excel. Tools such as Canvas offer a more efficient process to business managers as they look to compete in an environment affected by an increasingly diverse variety of factors. Tools such as ML should not be locked away from employees that are not part of the technology staff. History has shown that the democratization of technology (think of the journey from centralized computing to the PC) has unleashed innovation and increased productivity.