Using deep learning to analyze IoT data to predict product issues
"We were able to integrate Databricks with Tensorflow and Keras in a few clicks. And a 30+ node cluster would be up in less than four minutes."
Lennox International is an intercontinental provider of climate control products for the heating, ventilation, air conditioning, and refrigeration markets.
The Challenges
The sheer scale of their dataset did not allow them to process the data at the speed they required. As a result, they were forced to sample data from 15-20 devices and build models off of that subset of data
Their data teams have various skill sets (analysts, engineers, data scientists) and worked in silos which impeded productivity
Their data science team spent 80% of their day on data appropriation, data engineering, feature extraction, feature engineering, limiting their ability to train better and more accurate models
The Solution
Databricks provides Lennox International with a unified analytics platform on Microsoft Azure that simplifies data engineering and accelerates data science productivity.
Provides collaboration between team members
Automation and cluster management eliminates the need to tune and maintain Apache Spark™
Average time to start a cluster – 4 mins
Integrated H2O Sparkling Water, H20 Driverless AI, TensorFlow through GUI, and PowerBI
