MLXpress - Self Service Machine Learning at Scale
Predictive Modeling Platform
MLXpress enables a highly agile way of doing Data Science & building predictive models that’s fundamentally going to change the way our client do data science. It’s a solution to democratize Machine learning in an organization by allowing even non-data scientists build predictive models by providing data and their business objective.
MLXpress is integrated with our Data Lake solution – DLXpress, and helps in enabling everyone in an organization to build predictive models as per their business needs without being a Data Scientist. With the self-service UI, a user follows following steps:
Import Data as file or specify table in DLXpress that needs to act as base table.
Use SQL for performing any feature engineering
Visualize the data using Drag and Drop Self Service BI capabilities to understand the data
Analyze auto-computed statistics of data
Define Target Variable and attributes to be used for predicting the target
Select algorithms and related parameters
Submit to AutoML pipeline (powered by H2O.ai and DLXpress) to build best predictive models
View model leadership board to understand different models being evaluated their performance and interpret them on the same UI.
Download the best model
Solution Need
It’s a common challenge in industry that a Data Lake is build, but making the data & analytics still remain unlocked because of introduction of new technologies which users may not be aware of. MLXpress is Coforge’s solution for unlocking Machine learning capabilities available to non-data scientists with compute and scalable power of a Data Lake underneath.
Features
Built using 100% open source technologies
Benefits
Simplified way of building predictive models
Scaling Machine learning with compute power of Hadoop and Spark
Making machine process highly agile.
Technology Stack
R, Shiny, H2O.AI, Lime, GGPLOTLY, D3, DLXpress (for data onboarding)