Google Cloud Predicts World Cup Winners

The 2010 Football (‘Soccer’) World Cup had many wild predictions from all over the world. Yet, Paul the Octopus had made some very accurate predictions that brought him international fame overnight. Too bad for us, Paul is no more. So how do we still get our friendly water-cooler bets on the World Cup?

It seems that now Google’s Cloud Platform has come to replace Paul the Octopus. In Google’s annual developer­ focused conference Google I/O 2014, Jordan Tigani, a Google Engineer accurately predicted the outcomes of all the Round of 16 (video below). Now, how did Google manage to do that?

How does Google predict the World Cup?

Google’s Cloud Platform used a model that was built using touch-­by-­touch data from Opta, a high definition sports data collection company that is covering the World Cup. It used the data from many soccer leagues where World Cup players play and examined their behavioral patterns to predict their performance in subsequent games. It also utilized a number of Google services including: Google Cloud Dataflow, to ingest the data, Google BigQuery to build derived features, and Google Compute Engine to crunch the data. The Google Prediction API was also used in this model.

So what really is Google Cloud Platform?

Now, let’s understand Google’s Cloud Platform in detail. Google Cloud Platform is a family of products and services offered by Google, each including a web interface and a command line tool, along with additional programming interfaces and a RESTful API. Some of the components of the Google Cloud Platform are:

  • Google Compute Engine

    Is the Infrastructure as a Service (IaaS) service that provides virtual machines that run on Google infrastructure. However, all instances on Google Cloud Platform are linux based such as RedHat, Debian, CentOS etc. It also offers exclusive features such as enhanced security, global load balancing and pay­per­use servers and storage.

  • Google App Engine:

    Is the Platform as a Service (PaaS) service that enables users to develop and host web applications in Google’s huge data centers. It provides exclusive features such as free service for minimum resources and bulk downloading. It’s database uses a SQL­like syntax called GQL.

  • Google Cloud Storage

    Is the online storage web service that helps user to store and access their files on Google’s storage servers. It provides some phenomenal features such as Interoperability, resume-able uploads, and access control.

The other platform services include Cloud SQL, Cloud Datastore, Cloud DNS, Hadoop on Google Cloud Platform, BigQuery, Translate API etc.

What are the other real world applications of the Google Prediction API?

Imagine if you were a day trader (one who buys and sells stock many times daily to make an average gain), you would want to know how the stock would be trending in the next 10 minutes. Right?

Google’s Cloud Platform prediction model might just be the right thing for you. However, predicting the moving average of a stock takes in more than just the past patterns. Imagine calculating a person’s credit score using this, to predict the likelihood of his future payments. Or imagine predicting the currency exchange rates and the international flight prices.

One new start-up actually built their entire business on the Google Prediction API and other Google Cloud Platform Services.  Pondera Solutions, a fraud ­detection company built their services on Google’s cloud to provide FDaaS (Fraud detection as a Service).

This jump in innovation opens up endless possibilities for how the cloud and big data will shape every aspect of our day to day lives.