2014 is already turning out to be a great year for BigML. In the first few months of 2014, BigML announced the availability of the Winter Release, reached the 1-million-predictive-models milestone, became a Tableau Software technology partner and launched the open sunburst feature. And more exciting features are coming soon to the BigML platform.
An interactive SunBurst visualization can now be embedded in blog posts and web pages using a snippet of HTML code.
In January, ProgrammableWeb reported that BigML had announced the availability of the 2014 Winter Release, which includes new features and improvements that boost predictive modeling. The release also includes Flatline, a new Lisp-like language created and developed by BigML that comes with the BigML API. Flatline can be used to to transform the REST resources programmatically and allows for a new paradigm that BigML calls Programmatic Machine Learning.
The million-model mark
At the end of February, BigML reached a major milestone: More than 1 million predictive models have been created with the BigML platform—400,000 of those predictive models had been created in the two months prior to the announcement.
Image Credit: BigML.
About one year ago, BigML had reached the milestone of 10,000 predictive models created with the platform. Francisco J. Martin, BigML co-founder and CEO, said to ProgrammableWeb about the 1 million milestone: "A million models isn't cool. You know what's cool? A billion models. That's our next milestone."
Tableau software partnership
Earlier this week, BigML announced that the company had become a Tableau Software technology partner which will allow the analytics and visualization features of both platforms to be combined in a number of ways.
A new export feature has already been launched which allows BigML models to be exported directly to the Tableau platform as a calculated field. Tableau users can now interact with BigML models just as they would any other Tableau field.
Open SunBurst feature launched
SunBurst tree visualizations were introduced in mid-2013 as an alternative method of visualizing decision trees. A decision tree in machine learning is a model that predicts a target variable value based on multiple input variables. It is also used to create a simple representation for classifying examples.
In January, BigML announced the launch of the Open SunBurst feature, which makes it possible to embed a smaller version of an interactive SunBurst visualization into a blog post or Web page using a snippet of HTML code.
BigML API webinar
On March 11 from 10 a.m to 11 a.m. PDT, BigML will present a free webinar "Building Predictive Apps with BigML's API." The webinar will cover topics that help developers build machine learning predictive applications using the BigML API. Webinar topics will include (but will not be limited to):
- BigML API introduction
- Summary of bindings in other languages such as Java, Node.js, Clojure
- Dataset transformations
- Machine learning strategies: Covariate shift detection, boosting, smart feature selection.
Developers, data scientists and others interested in viewing the webinar can register to attend on the BigML website.
More exciting features coming soon
BigML is working on a lot of new features and improvements that will be coming out soon. Martin provided ProgrammableWeb with a few insights regarding the BigML road map. The company is currently working on three main areas: Extending the BigML API, a new clustering algorithm and organizations.
The company is working on extending the BigML API with a language to implement high-level machine learning strategies. As Martin explains it to ProgrammableWeb:
This language will allow our users to implement sophisticated machine learning templates in the cloud either programmatically or just in one click. This language together with our current transformational language for datasets will elevate programmatic machine learning to another dimension.
Another feature now in development is a new clustering algorithm that is the company's first take on unsupervised learning. The clustering algorithm, along with the current supervised algorithm, "will help solve many other problems that our customers are facing with their data," says Martin.
Organizations is a new feature that BigML is working on that will allow users to have multiple dashboards and collaborate with other co-coworkers. The BigML organizations feature will work somewhat like GitHub organizations.
2014 is shaping up to be a banner year for BigML, with more features coming soon. To learn more about the BigML platform, visit BigML.com.
By Janet Wagner. Janet is a data journalist and full stack developer based in Toledo, Ohio. Her focus revolves around APIs, open data, data visualization and data-driven journalism. Follow her on Twitter, Google+ and LinkedIn.