BigML Fall 2013 Release!

Posted by

BigML is excited to announce our Fall 2013 Release—the largest leap in functionality that BigML has taken since our service came out of Beta late last year.

Headlining the Fall 2013 Release are the following features:


  • Advanced Text Analysis: Have you ever wanted to create predictive models using a mix of structured and unstructured data but weren’t able to find a package or service available to do so?  Only BigML allows you to process multiple text fields, alongside numerical, categorical, date and time fields in a user friendly service—enabling you to spot relationships between text content and other properties of your data.  For example, wouldn’t it be good to know what keywords from your social media feed are resulting in the most social media shares?   


  • Microsoft Excel® Export:  You can now export your BigML models and make them actionable on Microsoft Excel spreadsheets. This allows you and/or your co-workers to run models “on the go” and it’s also great for sharing models in an offline fashion with clients and colleagues.


  • Multi-label Classification:  Sometimes you need to assign multiple classes (labels) to your data. For example, what keywords best describe a text?  BigMLer, our command-line tool for Machine Learning, makes multi-label classification very easy:  in just one line you can generate a combination of models that will help you predict all the classes to which a new instance belongs.


  • BigML PredictServer: For users who demand lightning fast predictive capabilities, BigML is pleased to introduce BigML PredictServer—a dedicated server on AWS that can process thousands of predictions per second based on your trained BigML model or ensemble.  Available today in early access form, BigML PredictServer is a great solution for customers that need predictions in real-time or in large batches.

BigML confusion matrix

  • Confusion Matrix: A highly-requested enhancement to BigML’s Evaluations, BigML’s intuitive and interactive confusion matrix makes it easy for you to visualize the performance of your classification models and ensembles by quickly showing you the actual vs. predicted results highlighting false negatives and false positives. Also exportable to Microsoft Excel.

BigML Secret Links

  • Secret Links: Have you ever wanted to share a model with a colleague, but not post it in BigML’s Gallery? Now you can through a simple copy & paste of a secret link!  As always, your data is completely safe with us—the link provides access to the model, and nothing else.


  • Inline Field Editing: Add labels and descriptions to fields in your sources and datasets that will carry over to your models and predictions—this is great for making your data easier to understand and analyze.

In addition to the above 7 new features, BigML has made hundreds of improvements to the UI, workflow, and added thousands of lines of code to our API and backend.

The best news of all is that this functionality is available today. These features work with your pre-existing data sources but you might need to recreate your datasets to include the new text analysis.  Just log in to your account or fire up BigMLer and enjoy all of these new capabilities!  We’ve made it easy for you to see the new features—just look for the “New” icons  throughout the interface:New badge

We’ve also added some new pages and resources to our website.  Check out our new BigMLer, PredictServer and Training pages, and stay tuned for the first set of companies that will be added to our new Partners page.

And as always, please let us know what you think.  We love hearing from our users, and are happy to help you out with your Machine Learning challenges. Just email us and we’ll get back to you ASAP.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s