One of the popular uses for machine learning is helping organizations sift through data to find new patterns, giving business executives information that is actionable.
CData, a North Carolina-based provider of data access solutions, last week made its first foray into the open-source world with the creation of the CData ODBC Reader for TensorFlow. This experimental project was created to enable people who want to apply machine learning to their data with TensorFlow to pull data from any ODBC-compliant data source, according to Eric Madariaga, CMO of CData.
(Related: Developer uses TensorFlow to design AI that can play Mario Kart)
“We’re offering a data abstraction layer without the need for big server-side code,” he explained. “Users can pick and choose their data sources. We provide the interface between TensorFlow and whatever ODBC data source you plug in.”
In a statement announcing the reader, Madariaga wrote: “Our goal in the ODBC Reader for TensorFlow is to help streamline the way that organizations connect machine learning with real-time data, regardless of where the data resides. By doing so, we hope to facilitate the growth of [machine learning] integration, and ultimately make it more accessible to organizations of all sizes.
The reader connects with CData’s ODBC drivers, providing access to data sources such as AdWords, Azure, Cassandra, Couchbase, Excel, MongoDB, SalesForce, SharePoint and more, Madariaga wrote in the statement.
He told of an effort to use TensorFlow with weather data, to look at Google’s pay-per-click advertising and how the weather impacted the amount of clicks. “It’s the same idea with our drivers,” Madariaga said. “We can get into those places to get data, and users can do what they want with it” from a machine-learning perspective.
Many big organizations collect data from any number of sources, upon which they base business decisions. CData’s drivers enable users to pull data from those sources without needing companies to write specific code for each data source, or to live with a single data dump that then has to be sorted through. “People today are interested in individual connectivity to different data sources,” Madariaga said.