Predict the future with Google Prediction API

Ajay Ohri
Jul. 30 2012, 01:00PM EDT

Google has been a leader in providing innovative APIs as well as providing infrastructure and  software paradigms (including map-reduce) that have helped make the internet better. The enormous investment that Google has put in these infrastructure efforts is yet to pay off directly, but it has brought many developers a choice and flexibility between the usual options. Of course if computing shifts to the cloud, Google will reap a big dividend from both retail as well as enterprise software customers.

The Cloud APIs of Google are-

1) Google Prediction API for Machine Learning based Models

2) Google Storage API for Data Storage

3) Google Cloud SQL API for Relational Database (or MySQL in the Cloud)

4) Google BigQuery API for Big Data Querying

You can see the rest of the APIs created by Google using the API console at

The Google Prediction API is the machine learning API of this suite. It assumes data is already in Google Storage (which has a separate API), and you need to apply supervised learning to it based on the training dataset. What can we do with this API?

You can use the Google Prediction API in some of the following custom cases- assuming Google inc would have better machine learning than your own team (which is a reasonable and cost effective assumption)

  • Customer sentiment analysis - as in Social Media Analytics or viewing the responses to your marketing in real time
  • Spam detection- If you need to build spam detectors for your in-house (Exchange based) email system. An alternative is to just go on the Google Cloud for Gmail for Businesses, which may not be a feasible solution for legacy or confidentiality reasons.
  • Upsell opportunity and Churn analysis - This can help with using these models than pay for highly expensive software that analyzes your customer data.
  • Suspicious activity identification- Especially if you have tonnes of web and machine logs to go into and are a hapless overworked Information Technology member in a corporate team who has been tasked with additional tasks of information security.

There is a rich variety of client libraries that are supported by this API including .NET , Go , Google Web Toolkit , Java ,JavaScript ,Objective C ,PHP , Python , Ruby , Apps Script ,R  .

Pricing is awesome for the Google Prediction API compared to the costs of building your own analytics team or buying analytics software. For 10$ per project , you can do 10,000 training and prediction calls per month.

Want to predict the future? It is just an API call away!

(Google Cloud APIs -part 1)

Ajay Ohri is the author of R for Business Analytics and likes to write on Enterprise ,Cloud and Statistical APIs with an emphasis on interviews. Follow Ajay on Google+ and connect on LinkedIn

Comments

Comments(7)