Skytree will primarily appeal to large organizations with some experience in the use of machine learning technologies; and in fact Skytree positions itself as ‘The Machine Learning Company’. Its Infinity platform provides the tools for analysts and data scientists to create predictive models, in a manner that is both productive and effective. Productivity benefits come from the automation of many tedious tasks that typically require weeks of fine tuning, and the effectiveness of the resulting predictive models is due mainly to the extensive data exploration tools, model performance monitoring, and the fact that most of the algorithms have been designed ground up, specifically for big data analytics. It is hard to see why a large, sophisticated organization might not consider Skytree in its efforts to realize the benefits associated with decision automation.
Skytree is used in a diverse set of industries, but has found its most rapid uptake in financial services, insurance, banking and where very fast, and accurate models are needed – specifically in businesses that need to make predictions from large volumes of data, such as from sensors and devices (the emerging Internet of Things). The very fast algorithms and support for advanced hardware configurations makes it an ideal platform for businesses that need to remodel frequently, and where high volume data streams need timely processing.
Applications in financial services are numerous and include fraud detection, where prevention needs to be balanced with minimizing false positives. As the fraudsters become more sophisticated, so the predictive models have to become smarter and remodeled more frequently to catch changing behaviors. Skytree is used in financial services business for fraud detection, and such is the efficacy of the resulting models that fraud rates can see reductions of up to ten per cent.
Other applications include prescriptive maintenance, where maintenance costs in asset rich organizations can be reduced by pre-empting failure, and optimizing service schedules. These applications often need real-time monitoring and the ability to optimize in real-time. The rapid growth of intelligent sensors and devices is part of the emerging Internet of Things, which will demand the performance and efficacy of the solution offered by Skytree.
The Skytree Infinity analytics platform inherits many years of leading edge developments in machine learning algorithms, and retains some of the world’s foremost authorities as advisors. The net result of this is very fast execution, accommodation of big data, and the automation of many repetitive tasks which slow predictive model development and deployment.
The platform isn’t simply a toolkit for model creation however, it includes data exploration capabilities, model visualization tools, model monitoring and a variety of capabilities which support predictive model governance. This is essential in highly regulated industries, where regulators demand that models are well understood and documented.
A facility called AutoModel allows data scientists and analysts to create models without having to choose a method, or fine tune the dozens of parameters associated with many machine learning techniques (pruning parameters in decision trees for example). Skytree claims that automation will usually produce models which are at least as good as, and are often superior to those which have taken weeks to tune by hand. In any case, there is no obligation to use the automation features, and this is no black-box, but with time users gain confidence and use the automation feature to speed development.
Most machine learning methods are supported by Skytree, including gradient boosted trees, nearest neighbor, collaborative filtering, random decision trees, support vector machines, K-means, linear regression, kernel density estimation, single value decomposition, principle component analysis … and many others. Most big data platforms are also supported including Cloudera, Hortonowrks, MapR, Spark and YARN. Other data sources can also be used if needed.
The actual model building environment encompasses several interfaces to Java, Python and the Skytree Command Line Interface. A graphical interface is also provided by Skytree that supports a project orient approach to model building.
Creating predictive models is just one part of the model lifecycle. They have to be monitored for performance, documented, modified (when circumstances change), understood, and any other activity which ensures that models are fit for purpose, and deliver maximum advantage. Skytree majors on supporting many of these activities, providing automatic model documentation, an audit trail of modifications, graphics to aid model understanding, and others to provide details of performance. Interfaces to business intelligence products such as Tableau allow management to track the impact of predictive models on business performance.
Skytree is a rapidly growing, privately held company with several investors. It is headquartered in San Jose California.