Many people know Google first and foremost as a search engine company. But really it’s a machine-learning company, using data to make predictions that get incorporated into applications like search and advertising without people even realizing it.
Today Google is announcing in a blog post that people can now choose to apply its machine-learning savvy to Google Sheets, the company’s spreadsheet app, to make educated guesses and fill in blank cells.
This applied use of machine learning follows Microsoft’s recent announcement of a cloud-based service for that purpose, Azure Machine Learning.
Google has previously offered Autofill capability that could figure out obvious information like the day of the week, a date in a month, or a number in sequential order. Now, with the Smart Autofill add-on, Google is do much more with numbers and words by drawing on its Prediction API, an application programming interface that considers existing numbers or categories and then creates a model to make reasonable approximations.
“The best model is automatically chosen for your problem by finding the one with the smallest misclassification error (for categorical data) or root-mean-squared error (for numeric data) calculated by using cross-validation on the labeled (non-empty) set of examples,” software engineer Konstantin Davydov and research scientist Afshin Rostamizadeh wrote in the blog post.
The service provides an error rate, so you can see how certain Google is in making its predictions with Smart Autofill, Davydov and Rostamizadeh wrote.