We have covered Machine Learning APIs before, including BigML.com, Algorithms.io and others. We are now continuing the series in an interview with Simon Chan of Prediction.io which aims to further bring machine learning closer to the common man, in an easier interface. PredictionIO is an open source machine learning server for software developers to create predictive features, such as personalization, recommendation and content discovery.
PW- Describe the core idea of PredictionIO:
Simon- PredictionIO is an open source Machine Learning server for software developers to build smarter applications.
With PredictionIO, you can add the following predictive features to your applications instantly:
Predict user preferences
Offer personalized video, news, deals, ads and job openings
Help users to discover interesting events, documents, apps and restaurants
Provide impressive match-making services
We take care of scalability and streamline the data engineering process so developers can focus on building the core of their applications. PredictionIO can be installed on-premises or be easily launched as SaaS on cloud services like AWS.
Simon- PredictionIO is built with programmers in mind. The first uniqueness is that, unlike other black-box products, PredictionIO is 100% open source. We want to make sure that developers can understand the product and take control of it, if and when they want to. The other uniqueness is that PredictionIO comes with developer-friendly APIs and UIs to streamline the whole data engineering process, no prior machine learning knowledge is required. It is, on the other hand, flexible enough to allow developers and data scientists to add their own algorithms on top of PredictionIO.
PW- What are the various languages /SDKs that you support via Prediction.io? What were the factors that enabled you to choose these languages?
Simon- Developers can interact with PredictionIO through RESTful APIs. A number of SDKs are provided: Python SDK, Java and Android SDK, Ruby SDK, PHP SDK. Node.js SDK and iOS SDK are also being developed by the community.
PW- How do you intend to make your API more exciting for developers to create applications for?
Simon- We intend to make PredictionIO's API developer-friendly and intuitive. Developers can build powerful predictive features with just a few line of codes through SDKs.
PW- What kind of pricing strategy do you intend to follow? What are some of the potential partnerships you intend to do for Prediction.Io?
Simon- Like other open source products, we will offer enterprise support services in the near future.
PW-Name a real life case study where Prediction.Io led to substantial savings vis a vis other methods:
Simon- PredictionIO is empowering hundreds of different applications right now! One recent example is Clay.io, which is an app store for HTML5 games. Clay.io built a recommendation engine in hours rather than months by using PredictionIO.
PW-Describe your startup journey. What made you go this route?
Simon- With a programming background as a software engineer, I founded a startup that built a news aggregation Platform. Back then, it took our team a few months to build a recommendation engine that was not even scalable enough to handle millions of news articles. It was at that time that I realized there is a big gap between data science and software development.
So, after a number of years of PhD research in Machine Learning, I am eager to build a product that can solve this problem. We need to empower developers with a tool to develop predictive features and harness Machine Learning and Big Data productively. That tool is PredictionIO. I believe that developers will be a major driving-force in taking Machine Learning to the next level by building smarter applications.