Enterprise Analytics

Lionsgate enables Analysts, Data Scientists and Application Developers to transform BI and Analytics into intelligent applications.

Lionsgate Analytics provides business analysts, data scientists and developers a distributed machine learning platform capable of scaling from a single machine to a thousand+ node cluster. Features include:

Speedy integration and deployment across teams and applications

Effortlessly deploy machine learning models and predictive analytics as RESTful services using Java Script, Python and Scala.

Prep, Model, Test and Deploy

Big data IDE, modeling and ETL tooling for data scientists and developers, a powerful integrated Analyst Notebook for business analysts and a java script development framework for developers.

Data Science as a Service

We work with clients either to build, operate and train data science teams or to provide data scientists as a service on a project by project basis.

Monitoring and Management

Programmatically monitor performance of services and models and perform operations like service rollbacks if a new service breaks. Monitor the operational health of your cluster and the load and performance of your services in real-time.

Hybrid Cloud Hosting

Deploy on premise, in the cloud or both, your choice.

Advanced Analytics

Lionsgate’s distributed machine learning capability is designed to enable analysts to wrestle information out of enterprise scale datasets.

Deep Learning / Fraud Detection

  • Use cases include pattern recognition in fraud, financial services and natural language.
  • Uncover the hidden linear and non-linear relationships between inputs and outputs by implementing deep learning methods.
  • Fine tune models using different radial basis functions and propagation methods.

Bayesian Learning / Supply Chain

  • Bayesian approaches find broad applicability in the supply chain management domain where the understanding of how disruptions impact the entire network are the focus.
  • Bayesian learning techniques allow model evolution in response to new observations.
  • Models are periodically refitted to account for errors/residuals observed in the system.
  • Through this refitting the latest information is propagated across the network and the static parameters on the relevant nodes are updated to reflect the changes.

Variable Reduction / CRM

  • Widely used in CRM where customers can be effectively grouped by relatively small number of key variables which may reflect demographics or some preferences.
  • Reduce the dimensionality of your data using powerful clustering models.
  • Gives the business a more intuitive view of the data and allows decision makers to focus on what is important.
  • Discriminant analysis and principal component allows for direct ranking of the variables in terms of their impact on the dependent variable of interest.

Random Forests / Next Best Action

  • These methods are widely used in “next best action” engines where an instant reaction to a customer interaction is critical. They are similarly used in customer churn models.
  • Once the model has been fitted using a training set these models facilitate real-time decision making on the classification of new unseen data and hence provide the ability to respond/react instantly.
  • Bootstrapping and other ensemble techniques fine tune decision/classification trees and random forests can be used to explore the deeper recesses of the data.