Automated Machine Learning Using Spark Mllib to Improve Customer Experience

Slides PDF Video

[24]7 is a predictive analytics company focused on improving customer experience for Fortune 100 companies. We capture data about 2.5B customer interactions every year and we use this data to build machine learning models that predict the customer intent across multiple channels – online, voice and chat. We use Graphx to connect visitor information across different channels. In the past we used traditional tool set like R and SciKit for model building and have been limited by their scale. We have chosen Spark and Mllib for automated feature engineering and machine learning. This talk will cover our architecture, the challenges we faced and how we solved them and how we leverage Twitter’s Algebird abstract algebra on top of Spark.

Photo of Sourabh Chaki

About Sourabh

Sourabh is working as lead engineer in [24]7 data team. He has developed an automated machine learning and analytics platform from scratch on top of Hadoop and Spark and deployed the same on [24]7 private cloud.