PredictionIO's $2.5M money bag suggests open-source is right for machine learning

PredictionIO, a startup that has crafted an open-source program to let developers add machine-learning smarts to their applications, might just be setting the tone for the next wave in data technology.

More than 4,000 people have shown interest in the project by giving it a star on GitHub, hinting at its merit among open-source initiatives. Now investors have thrown $2.5 million at PredictionIO so that development and commercial support of the project can continue.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":1507343,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,enterprise,entrepreneur,","session":"D"}']

AlchemyAPI, Skytree, Wise.io, and other startups with easy-to-use machine learning services or application programming interfaces have also gotten funding in the recent past. But PredictionIO stands out from those, and not just because it’s free.

“In our own experience as developers, we really hate black-box solutions, especially when it comes to developer tools,” Simon Chan, PredictionIO’s chief executive and a co-founder, told VentureBeat. “It’s actually more about flexibility and customizability.”

One could attribute the same advantages to some of the data infrastructure tools that underly an increasing number of applications today.

MySQL provided commercial support for its synonymous open-source relational database and rode into the sunset when Sun Microsystems acquired the company for $1 billion in 2008. More recently, MongoDB has grown into an IPO-bound company as it pushes the open-source MongoDB NoSQL database. And investors have poured big sums of money into Cloudera, Hortonworks, and MapR as they fight hard to commercialize open-source Hadoop tools for storing, processing, and analyzing lots of different kinds of data.

So it does seem clear that open-source is important in the data world. And perhaps when it comes to machine learning, PredictionIO will become a standard.

Developers can run PredictionIO by downloading it onto their computers, or they can run a version of it on the Amazon Web Services public cloud. A user interface lets developers determine which algorithm would be the best fit for a particular application.

The startup could base the business on providing commercial support for companies’ use of the open-source software. But Chan probably won’t stop there; he’s been considering an enterprise version of the software.

PredictionIO is written in Scala, unlike some existing open-source big data tools, such as Java-based Apache Mahout. That could make a difference in how many people end up using the software. Perhaps it could go well beyond specially trained data scientists.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":1507343,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,enterprise,entrepreneur,","session":"D"}']

“We’re a tool for designers and for developers,” Chan said.

Quest Venture Partners led the round in PredictionIO. Azure Capital Partners, CrunchFund, the Stanford-StartX Fund, and Kima Ventures also participated.

PredictionIO started a year and a half ago and is based in Palo Alto, Calif., in StartX office space. Five people work for the startup.