BigML is a cloud based machine learning platform with an easy to use graphical interface. It also provides simple mechanisms to incorporate predictive models into production applications through its REST API. The platform combines supervised learning (to build predictive models), unsupervised learning (to understand behavior), anomaly detection (used in fraud detection), data visualization tools (scatter-plots and Sunburst diagrams) and many mechanisms for exploring data. The modest pricing will make it attractive to medium and large businesses who want the benefits associated with machine learning without large upfront costs and implementation delays. BigML is a pragmatic, low cost, easy to use platform for building powerful predictive models.
Since its general availability in 2012 BigML has matured into a user friendly, highly functional machine learning platform used by large and medium size businesses as well as application developers. It avoids the unnecessary complexity associated with many alternatives, and at the same time provides ample mechanisms to build powerful predictive models, explore data and visualize it for greater understanding. Resulting models can be exported in various formats and languages, making it easy to embed them in production systems. The main benefits associated with BigML are speed of deployment, modest costs, flexibility, scalability, ease of use, and inbuilt support, training and consultancy.
BigML has found application in a diverse range of businesses ranging from pharmaceutical and financial services to law firms. The in-built anomaly detection algorithms are ideal for fraud detection, and as such find use in financial services firms. The clustering methods are ideal for exploring market behavior and any data where group behavior is predicated. A variety of businesses have benefited from the sophisticated supervised learning supported by BigML and predictive models have been built to predict customer churn, project costs and a variety of other business behaviours. BigML supplies a gallery of predictive model ‘templates’ addressing a wide diversity of business problems.
As a cloud based resource BigML is accessed via the web browser. Data can be loaded via a variety of mechanisms including direct access to cloud based resources (e.g. Amazon S3). There are a rich variety of data preparation tools available for cleansing, normalization, transformation, feature engineering, and BigML will automatically recognize and format data according to its type (categorical, date, text, numeric). Metadata relating to individual attributes is displayed, including a very useful distribution plot. Unsupervised learning is based on clustering (K-means or G-means), and finds application in customer segmentation, fraud detection, object verification and many other applications. The supervised learning algorithms employ CART tree like classifiers and decision forests. Credit scoring, churn prevention, predictive maintenance, recommender systems are just a few of the typical uses. The anomaly detection isolates and ranks anomalies and is used for behavioral authentication, data cleaning, intrusion detection and even video surveillance.
The user environment is driven by a drag-and-drop type interface, producing informative, interactive graphics, and the ability to drill down into data and model details as required. Model evaluation is straightforward with splits in training and test data, and confusion matrix identification of false positives and negatives. Resulting models can be exported into Java, Python NOde.js and Ruby code as well as Tableau or Excel format.
Subscription pricing is modest starting at just US$30 per month for individual users with modest needs. Enterprise packages start at US$2500 per month and include support, assistance and training. For those with security and privacy requirements BigML also offers private cloud deployments – either as a single-tenant cloud solution or as an on-premise implementation.